CEMDAR PWS - Informational OCT2026.pdf

Performance Work Statement (PWS)

Research and Development for Cost Estimating Modeling

Data Analytics and Research (CEMDAR) Development Effort for

Air Force Cost Analysis Agency (AFCAA)

October 2026 i

Table of Contents

INTRODUCTION AND PURPOSE ----------------------------------------------------------------------- 1

TASKS AND TASK REQUIREMENTS ------------------------------------------------------------------ 2 Task 1 (Enabling task) – Obtaining Permissions to Access Data Sources ------------------------ 3 Task 2 (Enabling task) – Develop a Plan of Action and Milestones (POA&M) ----------------- 3 Task 3 (Core task) – Development and Continuous Improvement of Data Pipelines for

Enterprise Cost and Technical Analysis -------------------------------------------------------- 5 Task 4 (Core task) – Data-Driven Analysis/Studies, Model Development and Application

Deployment ---------------------------------------------------------------------------------------- 12 Task 5 (Core task) – Contracts Database (Option CLIN) ------------------------------------------- 16 Task 6 – Surge Support (RDT&E, Option CLIN) --------------------------------------------------- 17

KNOWLEDGE, SKILLS, AND ABILITIES (KSA) --------------------------------------------------- 18 General Requirements------------------------------------------------------------------------------------ 18 Program Manager ----------------------------------------------------------------------------------------- 18 Chief Cost Data Scientist -------------------------------------------------------------------------------- 18 Expert Cost Estimator/Analyst ------------------------------------------------------------------------- 19 Senior Cost Estimator/Analyst -------------------------------------------------------------------------- 20 Journeyman Cost Estimator ----------------------------------------------------------------------------- 22 Junior Cost Estimator ------------------------------------------------------------------------------------ 22 Expert Data Scientist ------------------------------------------------------------------------------------- 23 Senior Data Scientist ------------------------------------------------------------------------------------- 23 Journeyman Data Scientist ------------------------------------------------------------------------------ 24 Junior Data Scientist ------------------------------------------------------------------------------------- 24 CSDR Subject Matter Expert -------------------------------------- Error! Bookmark not defined.

On-Demand Subject Matter Expert (SME) ----------------------------------------------------------- 25 Key Positions ---------------------------------------------------------------------------------------------- 25

SERVICE SUMMARY ------------------------------------------------------------------------------------- 25

DELIVERABLES -------------------------------------------------------------------------------------------- 27 Status Reports --------------------------------------------------------------------------------------------- 29 Associate Contractor Interface -------------------------------------------------------------------------- 30 Data Rights ------------------------------------------------------------------------------------------------ 30

EQUIPMENT, MATERIAL, INFORMATION, OR SERVICES ------------------------------------ 32 Equipment, Material, Information, or Services ------------------------------------------------------- 32 Contractor Acquired Property--------------------------------------------------------------------------- 32

GENERAL INFORMATION ------------------------------------------------------------------------------ 32 ii

Duty Hours ------------------------------------------------------------------------------------------------ 32 Place of Performance ------------------------------------------------------------------------------------- 33 Kickoff Meeting/Orientation Meeting ----------------------------------------------------------------- 33 Quality Control. ------------------------------------------------------------------------------------------- 34 Contractor Manpower Reporting Application (CMRA) -------------------------------------------- 34 Travel and Other Direct Costs -------------------------------------------------------------------------- 34 Employee Accountability and Turnover -------------------------------------------------------------- 35 Miscellaneous Paragraphs. ------------------------------------------------------------------------------ 36 Acronyms -------------------------------------------------------------------------------------------------- 39 Applicable Publications ---------------------------------------------------------------------------------- 42 Notional Plan of Action and Milestones -------------------------------------------------------------- 45 Metadata/Key Word Example -------------------------------------------------------------------------- 46 Data Fidelity Scorecard Example ---------------------------------------------------------------------- 47

APPENDIX A: EXAMPLE LIST OF PROJECTS ----------------------------------------------------- 50

PURPOSE AND SCOPE ---------------------------------------------------------------------------------- 54

CORE DATA PIPELINES ------------------------------------------------------------------------------- 54

ENTERPRISE APPLICATIONS ----------------------------------------------------------------------- 56

AI/ML DATA INFRASTRUCTURE ------------------------------------------------------------------- 57

CROSS-CUTTING EFFORTS -------------------------------------------------------------------------- 57

NOTES -------------------------------------------------------------------------------------------------------- 58

INTRODUCTION AND PURPOSE 1

The Department of the Air Force (DAF) has identified numerous gaps in the databases, methods, 2 and tools it regularly uses for cost estimating and analysis and has committed to improving the 3 efficiency and effectiveness of its cost estimating capabilities. This requirement invests in 4 strengthening, accelerating and elevating cost data and analytic capabilities using the latest 5 technologies to enable faster, more efficient, higher quality, and higher volumes of analytic 6 decision support from the DAF cost community. It shifts the focus from desktop-based tools to 7 cloud-based platforms; from stand-alone commodity-based datasets to integrated and relational 8 databases containing programmatic, cost, technical, and schedule information; from static 9 spreadsheets/reports to dynamic analysis and models connected to data pipelines; and from low-10 volume cost data modeling to highly scalable data collection, curation, analysis and models. 11

The objectives include (1) improving the quality and amount of data collected and available to 12 cost estimators, (2) building centralized/ enterprise relational and integrated cost, schedule, and 13 technical databases using data science best practices, (3) building the capability to optimize cost 14 estimators’ use of the larger amounts of data collected for cost estimating, and (4) providing 15 centrally available and authoritative datasets and to develop cost estimating relationships, 16 schedule estimating relationships, technical estimating relationships, benchmark tables, 17 dashboards/visualizations, cost models, and other tools that Air Force and Space Force cost 18 analysts can use to develop detailed, robust, and accurate cost estimates. The goal is to not only 19 improve the timeliness and quality of program life-cycle cost estimates for budgeting, 20 programming, and acquisition decisions but to enhance the capability to perform quick turn high-21 level force design and alternative capability analysis as well as lower-level program element 22 trade-off analysis. 23

Thorough and current cost datasets and databases are the cornerstones to the Air Force and Space 24

Force continuously improving the accuracy and consistency of its acquisition program cost 25 estimates required by statute (e.g., Title 10, United States Code § 3222, 3225, 4251, 4252, 4253, 26

4271, and 4328) and regulation (e.g., DoD Instruction (DoDI) 5000.02, DoDI 5000.73, DoDI 27

5000.75, DoDI 5000.80, DoDI 5000.85, DoDI 5000.87, and Air Force Instruction (AFI) 65-508) 28 in support of DAF and Department of War (DoW) acquisition decisions. Moreover, these 29 integrated datasets and databases as well as the accompanying analysis should serve as the basis 30 to what the DAF cost community provides during the Strategy, Planning, Programming, 31

Budgeting, and Execution processes. These analytically based cost estimates must be robust 32 enough to identify available budget trade-space and what the near-term risks are as decision-33 makers balance competing priorities as they aim to deliver critical out-year capabilities to pace 34 the threat. 35

Building credible and defensible cost estimates to assess program affordability and support 36 budget development starts with complete and accurate datasets, sound estimating methods, 37 models, and enterprise tools. This effortfocuses on primarily DAF requirements, but the 38 solutions will be made available to benefit the government cost community, including the Office 39 of the Secretary of Defense (OSD) Cost Assessment and Program Evaluation (CAPE) and other 40 service cost agencies, and help it meet its growing demands. 41

TASKS AND TASK REQUIREMENTS 42

The contractor will develop enterprise datasets, methods and tools for use by cost estimators to 43 meet the AFCAA Cost Estimating Modeling Data Analytics and Research (CEMDAR) 44 objectives of an up-to-date, searchable, user-friendly, centrally available data and method 45 repository that enables a high volume and timely delivery of quality, effective cost estimates. 46

The contractor shall adhere to all scheduled timelines, milestones, delivery schedules, and 47 administrative requirements to include scheduled on-time delivery of reports, data products, 48 billing invoices, staffing of personnel, and action item completion. The contractor shall provide 49 the necessary labor, supplies, equipment, and supervision to fulfill this effort as described in the 50 following tasks and subtasks as outlined below. 51

The contractor will systematically and continuously document DAF and joint service cost 52 estimating data requirements, then collect and store the required data, and provide analyses, 53 methods, tools, applications, and models for use by cost estimators. Data will be collected in a 54 manner that ensures complete and accurate documentation and based on the government’s 55 approval, as well as shared within the AFCAA’s cloud-based platform(s) of choice (an 56 appropriately designated platform/system for these products). Currently, platforms utilized are 57 within DAF Data Fabric and include DAF Data Visible, Accessible, Understandable, Linked and 58

Trusted (VAULT) data platform as well as DAF Envision, Other utilized systems could include, 59 for example, GenAI.mil and CAPE’s CADE system. The current architecture is stable but could 60 change over time, and the contractor should be flexible to evolve with any changes to DAF Data 61

Fabric. 62

The contractor shall be responsible for ensuring the accuracy, timeliness, and completion of all 63 tasks under this effort. Replacement or substitute personnel for vacancies shall be filled with 64 qualified personnel within fourteen (14) calendar days of vacancy, in accordance with Table 1: 65

Performance ThresholdsTable 1: Performance Thresholds in Section IV, unless approved in writing 66 or otherwise directed by the Contracting Officer (CO). 67

Meetings will include a kickoff meeting within ten (10) business days after the start of the 68 contract award and each exercised option; meetings as needed between the contractor’s Program 69

Manager and AFCAA’s Program Manager to track performance, resolve issues, and ensure 70 compliance with the Performance Work Statement (PWS); status meetings as needed between 71 the contractor’s project leads and the AFCAA technical leads; and quarterly formal reviews to 72 include a final review about 30 calendar days before the end of the period of performance 73 between the contractor and the entire AFCAA government team to include the AFCAA 74

Technical Director, AFCAA Program Manager, and the AFCAA technical leads. All meetings 75 shall use Microsoft Teams as the default choice. Delivering the agenda and recapping the 76 meeting, as outlined in Table 2: Task Order Deliverables in Section VTable 2: Task Order 77

Deliverables, is the contractor’s responsibility for the kickoff meeting and formal reviews, 78 including the final review. 79

The contractor shall provide the COR with all passwords, data collected, processes used, and any 80 other information obtained throughout the contract’s period of performance. The COR will meet 81 with the contractor within 7 business days prior to the end of the contract to ensure all 82 information is received. Once confirmed, the contractor will send a letter to the COR confirming 83 that the data rights requirements (paragraph 0) were completed not later than five business days 84 prior to the end of the contract. 85

Task 1 (Enabling task) – Obtaining Permissions to Access Data Sources 86

Successful performance of this requirement depends on the contractor working proactively with 87 the government to obtain permissions to access government held data. Much of the data in the 88

Cost Assessment Data Enterprise (CADE) repository is from defense contractors and major 89 subcontractors. Each considers its data as business sensitive and subsequently has marked it, or 90 considered it, as proprietary. This requires protecting this information from public disclosure. 91

Since the contractor team while providing these services will be provided or have access to 92 nonpublic information, each person on the contractor and subcontractor team working this effort 93 shall sign a non-disclosure agreements (NDA) with AFCAA (Error! Reference source not 94 found.) and an NDA with OSD CAPE for CADE (Error! Reference source not found.). The 95 contractor and subcontractor team shall work proactively and sign, as required, NDAs with other 96 government organizations to support and fulfill data collection and task requirements. 97

The contractor will work proactively with AFCAA to identify NDA requirements for Original 98

Equipment Manufacturer (OEM) and major subcontractors. Should an OEM or major 99 subcontractor require an NDA with the contractor, then the contractor shall develop and 100 complete this before AFCAA provides access to this OEM’s or major subcontractor’s 101 government-held data. These NDAs must carry through the entire contract period of 102 performance, including all option contracts. Furthermore, the NDAs should flow equally across 103 the contractor and its subcontractors. These NDAs shall be coordinated directly between the 104 contractor and the OEMs within 30 days of identified need. If the contractor showsa diligent 105 effort to complete the NDA process with consistent follow up with OEMs and is still 106 unsuccessful, government will reassess and replan project accordingly. Subcontractors may or 107 may not be required to sign separate NDAs, dependent on the negotiated OEM agreements. 108

As part of its proposal and throughout the duration of this contract, the contractor shall identify 109 and notify the Contracting Officer of any Organizational Conflict of Interest (OCI) it or any of its 110 subcontractors has, or could have, with any OEM or major subcontractor who has submitted 111 reports held in CADE or the Earned Value Management Central Repository (EVMCR). 112

Moreover, the contractor shall notify the Contracting Officer and the AFCAA Program Manager 113 of any acquisition or merger involving itself, or its subcontractors and explain whether that 114 action will cause create an OCI. Similarly, the contractor will identify any other emerging event 115 or action that would create an OCI. 116

Task 2 (Enabling task) – Develop a Plan of Action and Milestones 117

(POA&M) 118

The contractor shall develop and submit, for government review and approval, a POA&M 119 describing the contractor’s approach to satisfy all tasks and elements for data ingestion and 120 curation, analysis, and other task requirements for the base period within 30 calendar days of 121 contract award and for each option year within 15 business days of the option year start. The 122

POA&M will be a living document that will be modified throughout the entire contract period of 123 performance, including all option periods. 124

a. As part of the kickoff meeting and at the start of the base period and each option 125 period, the contractor and AFCAA will meet to discuss POA&M development and 126 subtasks for inclusion. A draft POA&M shall be delivered for review and comment 127 not later than (NLT) 30 calendar days after contract performance begins (and within 128

15 business days after each option year begins). The contractor will work with the 129

AFCAA Technical Director, its Technical Leads, and its Program Manager to obtain 130 approval of a final POA&M. The POA&M will include, as a minimum: 131

b. A detailed description of the approach to completing each task (with element level 132 descriptions) and subtask. 133

c. Task and subtask activity and milestone schedule, including interim, draft, and final 134 milestones, products, and contract project deliverables and a description of risk 135 associated with completing the planned schedule. Interim milestones should include 136 any events or progress points that would require a go or no-go decision from the 137 government based on research to-date prior to moving ahead. The schedule will be 138 used for tracking purposes and should be included in eachStatus ReportsStatus 139

Reports. 140

d. The contractor’s plan should include sufficient lead time for the delivery of all project 141 deliverables to allow the government to review and provide feedback. The contractor 142 shall resolve all the government’s comments before any deliverable can be considered 143 final. This shall occur not later than ten business days prior to the end of the contract 144 base period and each contract option year end. Waiver can be requested if 145 government review not received in a timely manner 146

e. Each task and subtask will document approvals to collect classified data. This 147 includes tracking potential data aggregation or compilation as determined by program 148 classification guides. The contractor shall ensure the collected data are protected at 149 the appropriate level necessary to safeguard this information. The contractor shall 150 work closely with the AFCAA technical leads to gain approval to collect classified 151 data. 152

f. A management plan explaining the contractor’s approach to staffing, schedule 153 management, business management, and reporting. This plan will track and align to 154 both task and subtask activity and schedule0. The plan will be specific in the Chief 155

Cost Data Scientist and Program Manager (PM) role and activities. 156

g. Plans to format and share CEMDAR deliverables as working documents and to 157 finalize all products for upload onto AFCAA’s cloud-based platform(s) of choice (for 158 example, currently the DAF VAULT data platform and DAF Envision), the CADE 159 system, or any other designated repository. These deliverables will include, but are 160 not limited to, all datasets, databases, tools, models, reports, documents, manuals, 161

Data Item Descriptions, training materials, and dictionaries required to upload and 162 share the CEMDAR products. 163

The contractor shall be available to brief the POA&M to the government, along with any risk 164 areas associated with completing the planned schedule, to the AFCAA Technical Director, the 165

Program Manager, and the COR. The POA&M is a ‘living document’ and the contractor will 166 provide recommended changes/updates to the government at the quarterly meetings, or as 167 needed, and will be given direction and/or approvals by the government on the recommended 168 changes/updates to this document. 169

After the government sends its statement of intent to exercise its next option, AFCAA will 170 inform contractor of the specific requirements to be incorporated into the POA&M. The 171 contractor will respond to AFCAA’s requests to responds and update the POA&M. 172

Task 3 (Core task) – Development and Continuous Improvement of Data 173

Pipelines for Enterprise Cost and Technical Analysis 174

Task Objective 176

The contractor shall design, develop, enhance, and continuously improve a portfolio of 177 automated data pipelines to integrate disparate data sources into a unified data environment, 178 enabling enhanced cost, schedule, and technical analysis. The pipeline portfolio exists at varying 179 stages of maturity: some pipelines are fully operationalized and require ongoing enhancement 180 and continued development; others are in active development and require completion, testing, 181 and transition to operational status; and others have not yet been initiated and will require end-to-182 end design, development, and deployment. The contractor shall assess the current state of all 183 pipelines at the outset of each performance period and work with AFCAA to categorize each 184 pipeline by maturity level. The contractor shall then contine the development, enhancement, and 185 refinement of operational pipelines to advance their capability and extend coverage; complete, 186 validate, and operationalize in-development pipelines; and execute full lifecycle development for 187 new pipelines as prioritized by AFCAA. This work shall follow an iterative approach — 188 delivering working increments of capability, incorporating user feedback, and refining products 189 progressively rather than through a single delivery cycle. The POA&M shall reflect a strategic, 190 phased execution plan based on AFCAA-identified priorities, accounting for pipeline 191 dependencies, resource constraints, and the sequencing required to mature the overall CADOM 192 data environment. These pipelines will conform to the organization's defined data operating 193 model and support recurring and ad-hoc analytical needs, developed in AFCAA's platform of 194 choice and optimally utilizing available DAF technologies and toolsets. In addition to data 195 pipelines, this task includes the development and scaling of enterprise analytical tools and 196 applications — cross-commodity, centralized capabilities built on the CADOM curated data 197 environment and serving the broad cost estimating community — as further described under the 198

Enterprise Application and Tool Development subtask below. 199

Currently, systems utilized are withing the DAF Data Fabric and include DAF Data Visible, 200

Accessible, Understandable, Linked and Trusted (VAULT) data platform and Envision platform, 201 but this is subject to change over time as the DAF Data Fabric evolves. 202

Scope of Work 204

The contractor shall perform the following subtasks, as required by AFCAA in the approved 205

POA&M: 206

Cost Analysis Data Operating Model (CADOM) Data Architecture 207

Design/Support 208

• Collaborate with government stakeholders to assist in data architecture design/support 209 as required, to include: 210 o Assistance in conducting assessment of platforms, toolsets and assistance in 211 defining the optimal solution/tech stack meeting all AFCAA’s requirements 212 and considering all constraints. This also includes ensuring innovative 213 methods are utilized throughout the development of the CADOM 214 o Assistance in architecting solutions that meet unique government 215 requirements, such as data and applications that span various classification 216 environments, and spanning users that have different levels of access to data 217 o Assistance in maintaining the Standard Operating Procedures (SOPs) for the 218

CADOM 219

o Assistance in developing and incorporating optimal process user/Subject 220

Matter Expertise (SME) feedback and workflows 221 o Assistance in identifying and designing workflow integration points where 222

CADOM data pipelines and products can be connected to cost estimator 223 workflows to reduce manual data handling, improve data access, and 224 accelerate recurring analytical processes. This includes identifying repetitive 225 manual steps in current estimating workflows that could be automated or 226 streamlined through pipeline-connected tools, self-service data access, or 227 templated analytical workflows within the CADOM environment 228 o After contract award, contractor will work proactively with government to 229 design architecture and governance decisions, and collaboratively define and 230 distinguish between Level 1, Level 2, and Level 3 decisions. Level 1 decisions 231 can be made by the contractor with approval of their Chief Cost Data Scientist 232 and documented/communicated to the government. Level 2 decisions require 233 the written approval of the AFCAA technical lead, and Level 3 decisions 234 require the written approval of the AFCAA technical lead in coordination with 235 key stakeholders, to be defined by AFCAA. This allows for flexibility where 236 possible, while ensuring safeguards are in place for decisions that have major 237 impact and/or affect downstream products. 238

Data Source Identification and Assessment: 239

• Collaborate with government stakeholders to identify and document relevant data 240 sources across the organization to feed the data pipelines described below. Note that 241 data and data pipelines are expected to include (but are not limited to): Cost and 242

Schedule Data Reporting (CSDR), Earned Value Management (EVM), Selected 243

Acquisition Report (SAR/MSAR), Contracts, Budget, Air Force Total Ownership 244

Cost (AFTOC), etc. AFCAA will work with pipeline owners to allow access to data 245 sources as required, and as NDAs allow, to the contractor team. 246

• Conduct a comprehensive assessment of each data source, focusing on: 247 o Data Quality: Evaluate data accuracy, completeness, consistency, and fidelity 248 using defined metrics (e.g., leveraging a Data Fidelity Scorecard like the 249 example in reference g of this document). 250 o Data Normalization Requirements: Determine the necessary normalization 251 processes, including inflation/price escalation adjustments, Work Breakdown 252

Structure (WBS) alignment, and Cost Element Structure (CES) 253 standardization. Justify the chosen normalization methods. 254 o Data Availability and Access: Investigate data security classifications, 255 aggregation limitations, proprietary restrictions, access requirements, and 256 control information. Adhere to all applicable security policies and regulations. 257 o Data Source Characteristics: Document any other relevant database 258 characteristics (e.g., data formats, API availability, data update frequency) that 259 may affect data collection, management, and use. 260

Data Pipeline Design and Development: 262

• Design data pipelines that adhere to modern data architecture principles and leverage 263 appropriate technologies (e.g., Apache Spark, Python, R, SQL, Databricks, cloud-264 based data warehousing solutions). Technologies/tools utilized should be compatible 265 with current DAF Data Fabric and DoW Platforms (VAULT/Envision/GenAI.mil); 266 new tools can be suggested if compatible with government platform security 267 requirements and if their utilization does not introduce additional costs or any delay 268 or risks to the schedule as compared to utilizing technologies already existing on 269 these platforms. 270

• All pipeline code, configuration and infrastructure-as-code definitions shall be 271 maintained in a government-accessible version control repository (e.g. government 272 provided GitHub) designated by AFCAA. The repository shall include sufficient 273 documentation and dependency definitions for the government to independently 274 reproduce pipeline deployments from the repository contents. Commit history shall 275 be maintained. The contractor shall not maintain pipeline code exclusively in local or 276 contractor-controlled environments. 277

• Develop modular, scalable, and maintainable data pipelines for data extraction, 278 transformation, and loading (ETL) processes. 279

• Implement automated data quality checks and validation rules wherever possible and 280 optimal within the pipelines to ensure data accuracy and consistency. 281

• Develop error handling and logging mechanisms to facilitate pipeline monitoring and 282 troubleshooting. 283

• Maintenance, augmentation, and application of metadata schema- The contractor 284 shall maintain and augment the existing AFCAA metadata schema as new data 285 sources are onboarded and pipeline requirements evolve. Proposed changes to the 286 schema (additions, modifications, or deprecations of fields) require AFCAA technical 287 lead concurrence prior to implementation. The contractor shall apply metadata tags 288 consistent with approved schema to all data ingested, transformed, and curated 289 through Task 3 pipelines. Metadata tagging shall occur as an integrated step in each 290 pipeline. At minimum, each dataset passing through a pipeline will be tagged with the 291 applicable fields from the approved schema (e.g. program identifiers, commodity 292 type, WBS mapping, etc). The contractor shall develop and execute a quality 293 assurance process to verify metadata completeness and accuracy, reporting tagging 294 compliance rates quarterly. The metadata layer shall remain queryable such that users 295 can discover and filter available data across all pipelines from a single interface. 296

• 297

Data Pipeline Acceptance and Validation: 298

• Each Data pipeline shall undergo a formal acceptance process prior to being 299 considered operations. Acceptance criteria shall be defined collaboratively with 300

AFCAA at the design phase and documented in the POA&M for each pipeline. At a 301 minimum, acceptance requires : 1) successful end-to-end execution against pipeline 302 data; 2) automated data quality checks passing at or above thresholds as defined by 303

AFCAA; 3) documented data lineage from source to curated output 4) error handling 304 and altering mechanisms demonstrated to be functional and 5) government review 305 and written approval by the AFCAA technical lead. A pipeline shall not be reported 306 as complete in status reports until formal acceptance is documented. 307

Data Ingestion and Transformation: 309

• Ingest data from identified sources into a raw data storage area (e.g., a data lake). 310

• Perform data cleansing, normalization, and transformation operations to conform to 311 the organization's data model. 312

• Create curated data tables and views optimized for analytical queries and reporting. 313

• Implement data lineage tracking across all pipelines such that any curated data 314 element can be traced back to its original source record through each transformation, 315 normalization, and aggregation step applied. Lineage documentation shall identify, at 316 a minimum, the source system, extraction date, and each transformation applied — 317 including the business rule or method that governed the transformation. Lineage shall 318 be maintained as an integral component of the pipeline, updated automatically as 319 pipelines execute, and accessible to authorized AFCAA users through the CADOM 320 environment. Where data is derived from multiple sources or combined through joins 321 or aggregations, the lineage shall document all contributing sources and the logic 322 used to combine them. The contractor shall ensure that lineage information is 323 sufficient for an independent analyst to understand how any output value was 324 produced and to reproduce the result given access to the same source data. 325

Data Repository Development and Maintenance: 327

• Organize data tables and views into user-friendly, searchable data repositories (such 328 as a data catalog), leveraging existing available technologies wherever available, and 329 customizing solutions where available technologies are not present in the cloud-based 330 platform of choice. 331

• Ensure the data repositories are optimized for performance and scalability. 332

• Work with AFCAA to collaboratively design a data security model to protect 333 sensitive data, including role and/or attribute based access controls consistent with 334

NDA requirements established under task 1. This includes defining and document an 335 access control model for end-to-end data pipelines, accounting for OEM proprietary 336 restrictions, data aggregation limitations, and classification requirements. The access 337 control model shall be documented, maintained as part of CADOM SOPs, and 338 reviewed with AFCAA quarterly. The data security model will be implemented and 339 maintained and controlled by government personnel, Develop and maintain data 340 dictionaries and metadata documentation to facilitate data understanding and usage. 341

• Document and maintain data fidelity information (e.g. the data fidelity scorecard 342 example in reference g) to facilitate user data understanding and usage. 343

Enterprise Application and Tool Development 345

The contractor shall design, develop, and scale enterprise-level analytical tools and applications 346 that operate on the CADOM curated data environment and serve the broad cost estimating community 347 across commodities. Enterprise applications are centralized, cross-cutting capabilities — examples include 348 enterprise data discovery and catalog interfaces, cross-commodity benchmarking and visualization tools, 349 and enterprise estimating capabilities built on curated pipeline data — developed as prioritized by AFCAA 350 through the POA&M. 351

Task 3 / Task 4 boundary. An application falls under Task 3 when it (1) originates from an enterprise or 352 cross-cutting capability requirement rather than a single scoped use case, and (2) is intended for centralized, 353 multi-commodity use. Applications developed in direct support of a specific scoped user requirement — 354 typically a single commodity, program, or study, and not yet appropriate for a centralized solution — are 355 developed under Task 4. In coordination with the AFCAA technical lead, the contractor shall classify each 356 application effort against this boundary at the time of authorization and document the determination in the 357 applicable POA&M entry. 358

Promotion of Task 4 applications. A Task 4 application that demonstrates broad value across commodities 359 may be designated by AFCAA for promotion to Task 3, where the contractor shall generalize, harden, and 360 scale the capability for enterprise, multi-commodity use. Promotion is a Level 3 decision under the decision 361 framework above and shall be authorized through the POA&M. As part of the enterprise build, and where 362 directed by AFCAA, the contractor shall support transition from the predecessor Task 4 application — 363 including data migration, user cutover, and orderly retirement (decommissioning) of the superseded tool — 364 and shall record retirement as a milestone in the POA&M. 365

Development discipline and operation. Enterprise applications shall follow the same acceptance, version 366 control, and repository discipline required for pipelines: all application source code, configuration, and 367 infrastructure-as-code definitions shall be maintained in the government-accessible repository with 368 sufficient documentation for the government to independently redeploy, and shall be deployed to the 369 government's platform of choice following established SOPs. Consistent with the CADOM operating 370 model, steady-state operation and hosting of deployed enterprise applications are performed by government 371 personnel; the contractor's role under this task is the development, enhancement, generalization, and 372 continuous improvement of these capabilities, not their steady-state operation. 373

Dependencies. Enterprise application work depends on the underlying Task 3 pipelines and on the metadata, 374 lineage, and data security models established elsewhere in this task. The contractor shall coordinate 375 enterprise application development with related Task 4 projects through the dependency matrix 376 (Deliverable 3). 377

Documentation and Training: 379

• Develop comprehensive and user-friendly documentation for each data pipeline, 380 including data source descriptions, ETL processes, data transformations, data quality 381 checks, and data lineage information. Integrate, wherever possible the documentation 382 for data pipelines within the CADOM architecture in an efficient way (i.e. GitHub, 383 markdown, tooltips, Wiki, etc). 384

• Develop and deliver role-based training tailored to distinct user audiences. At a 385 minimum, training shall address two roles: (1) cost analysts and estimators, focused 386 on data discovery, querying, interpretation, and use of analytical tools and 387 dashboards; and (2) data engineers and pipeline maintainers, focused on pipeline 388 architecture, code structure, configuration, deployment, troubleshooting, and 389 extension. Training content, depth, and format shall be appropriate to each audience. 390

• Provide detailed documentation for future enhancements and maintainers of the 391 developed pipelines. 392

• Create reusable training artifacts that persist beyond live training sessions and remain 393 accessible to current and future users. Reusable artifacts shall include, at a minimum: 394 recorded training sessions or walkthroughs, hands-on lab exercises or guided 395 examples using representative data, and interactive documentation (e.g., annotated 396 notebooks, wiki-based guides, or in-platform tutorials). Training artifacts shall be 397 stored in the Government-designated repository and maintained concurrent with 398 pipeline and product updates. 399

Continuous Improvement (Once data pipelines have been developed and deployed): 401

• Once a data pipeline is accepted, the contractor shall continue iterative development 402 and enhancement of the pipeline — incorporating user feedback, extending coverage, 403 and improving performance and data quality. Consistent with the CADOM operating 404 model, operation of accepted pipelines, including execution of scheduled data refresh 405 cycles at the frequency defined in the pipeline's acceptance documentation (e.g., 406 daily, weekly, monthly, or event-triggered), is performed by government personnel; 407 the contractor shall provide the automation, configuration, and documentation 408 necessary for the government to operate each pipeline. 409

• The contractor shall acknowledge reported defects or data quality issues in pipelines 411 under development or enhancement within 2 business days and provide a remediation 412 plan within 3 business days. 413

• Contractor shall stay abreast of new technologies and best practices in data 414 engineering identify areas for optimization, suggesting potential improvements to 415

AFCAA, where applicable. 416

Technology insertion and iterative development 417

The contractor shall employ an iterative development approach for pipeline design, development, 419 and deployment. Working increments of pipeline functionality shall be demonstrated to AFCAA 420 technical leads at intervals defined in the POA&M to enable early feedback and course 421 correction prior to formal deliverable milestones. The contractor is not required to deliver all 422 pipeline functionality in a single release; incremental delivery of working capability is expected 423 and encouraged. 424

The contractor shall continuously evaluate emerging technologies, tools, and platform 425 capabilities that may improve pipeline performance, maintainability, scalability, or cost-426 effectiveness. When the contractor identifies a technology insertion opportunity — including but 427 not limited to adoption of new tools, migration to updated platform services, replacement of 428 deprecated components, or integration of new DAF-provided capabilities — the contractor shall 429 submit a Technology Insertion Recommendation to the AFCAA technical lead. The 430 recommendation shall describe the proposed change, the expected benefit, the impact on existing 431 pipelines and deployed Task 4 products, the estimated level of effort, and any risks. AFCAA will 432 evaluate and approve, modify, or reject the recommendation. 433

Approved technology insertions shall be incorporated into the POA&M and executed within the 434 existing contract scope, provided the insertion does not materially alter the overall level of effort 435 or require capabilities outside the labor categories defined in this PWS. Technology insertions 436 that would materially change the contract scope or level of effort shall be identified to the 437

Contracting Officer for disposition. 438

The naming of specific technologies in this PWS (e.g., Apache Spark, Databricks, Python, R, 439

SQL) is illustrative and does not restrict the contractor to those technologies. The contractor may 440 propose alternative or additional technologies through the Technology Insertion 441

Recommendation process described above. Similarly, changes to the Government's cloud-based 442 platform of choice or associated toolsets shall not be considered out of scope; the contractor shall 443 adapt pipelines and products to platform changes within timelines coordinated with AFCAA. 444

AFCAA will evolve and adapt its Cost Analysis Data Operating Model (CADOM) over time to 446 adapt to changes/evolutions in available platforms, technologies and toolsets. The current 447

CADOM architecture is pictured in Figure 1 for reference. 448

Figure 1- Current Cost Analysis Data Operating Model (CADOM) Architecture, for reference purposes 452

Data Fidelity ScorecardData Fidelity ScorecardThe contractor shall support AFCAA in 453 collaborations with other cost organizations to identify data sources for collection efforts. These 454 include but are not limited to: the Air Force Life Cycle Management Center, the Space Systems 455

Command, the Air Force Nuclear Weapons Center, the Air Force Sustainment Centers, the Air 456

Force Rapid Sustainment Office, the Naval Cost Agency, the Naval Systems Commands, the 457

Deputy Assistant Secretary of the Army for Cost & Economics, the OSD CAPE, the Department 458 of Homeland Security, the National Aeronautics and Space Administration, the National 459

Reconnaissance Office, the Navy Strategic Systems Program Office, and the Missile Defense 460

Agency. 461

Task 4 (Core task) – Data-Driven Analysis/Studies, Model Development 462 and Application Deployment 463

Metadata/Key Word Example0Metadata/Key Word ExampleData Fidelity 465

Scorecard0Task Objective: 466

The contractor shall perform in-depth cost, schedule, and technical analyses and/or studies, 467 develop dynamic predictive models, custom applications, and insightful visualizations; and 468 deliver actionable recommendations to improve cost estimating methodologies and decision-469 making across the organization, as required by the approved POA&M. The goal of this task is to 470 provide data-driven insights, analysis, and tools tailored to AFCAA's needs. This task will 471 include projects that are intended to be fully deployed within the CADOM, projects that are 472 intended to be fully deployed outside the scope of the CADOM, and those that are a hybrid of 473 these two. This task shall ensure optimal delivery and integration of analysis/studies within the 474

CADOM ecosystem or as standalone deliverables, as determined by AFCAA. The contractor 475 shall leverage, to the greatest extent possible, the existing and future data pipelines developed 476 and delivered in Task 3 for Task 4 projects. The contractor shall assist the government in 477 identification of opportunities to migrate traditional studies/analysis to the CADOM and/or other 478 modern platforms/toolsets wherever synergies and/or efficiencies could be achieved. The 479 contractor shall develop and maintain a dependency matrix mapping Task 4 projects to their 480

Task 3 pipeline dependencies, where applicable (some Task 4 projects will not have any Task 3 481 pipeline dependencies). This matrix is identified in as Deliverable 3 in Table 4, Deliverables. 482

When a Task 4 project requires a pipeline that is planned but does not yet exist or requires 483 modifications to an existing pipeline for use, the contractor shall document that requirement in 484 the applicable project authorization document and coordinate with the AFCAA technical lead to 485 incorporate the pipeline work into the Task 3 schedule. When changes to a Task 3 pipeline might 486 affect deployed Task 4 products, the contractor shall perform impact analysis and notify affected 487 project leads prior to implementing the change. 488

Scope of Work: 489

The contractor shall develop use cases, dynamic analysis/predictive models, custom applications, 490 and conduct cost and technical studies as required per approved subtasks in the POA&M. The 491 subtasks in the POA&M, depending on the type of task, will include some combination of data 492 analysis, model development, application/tool development, and conducting targeted cost and/or 493 technical studies. Subtasks as defined in the POA&M could fall into one or more of these 494 categories as listed below. Each category is further defined below, along with potential 495 expectations and/or examples of what will be required (each subtask will be clearly defined 496 within POA&M). To facilitate planning and illustrate the scope of work under this task, a 497 representative list of example projects is provided in Appendix A. These projects are indicative 498 of the types of data-driven analyses, studies, and model development efforts that will be executed 499 annually for the base year and each subsequent option year. The Government will introduce new 500 projects each year. It is understood that project timelines will vary; some larger projects may 501 span several years, while others may be completed in less than a year. Furthermore, it is 502 acknowledged that for many of the studies undertaken, readily available data may not exist, and 503 data collection and normalization will be a key component of the research effort. 504

Data Analysis, Model Development, Cost, Schedule and/or Technical Studies: 505

• Conduct thorough data analysis and/or develop models using available data pipelines 506 and/or other data sources. 507

• Identify trends, patterns, and relationships within the data relevant to the identified 508 use cases, including but not limited to cost estimating benchmarks, methods, Cost 509

Estimating Relationships (CER), Schedule Estimating Relationships (SER), 510

Technical Estimating Relationships (TER), and other analysis/estimating approaches, 511 as required. 512

• Develop cost model architectures and/or models, that could be used as a template/ 513 starting point, early acquisition trades, etc for analyst developed Life Cycle Cost 514

Estimates. Note that development of any LCCEs is out of scope for this effort but instead this 515 is meant to capture efforts enabling this LCCE development. 516

• Employ appropriate statistical and/or machine learning techniques to build predictive 517 models that improve forecasting accuracy and risk assessment. 518

• Ensure all models are rigorously tested and validated using appropriate statistical 519 methods. 520

• Conduct targeted studies to explore specific concepts, commodities, and/or programs. 521

• Create interactive dashboards and visualizations that provide insights into key 522 performance indicators and trends. 523

Application and Tool Development: 525

Applications and tools developed under this task are those built in direct support of a specific scoped use 527 case or user requirement — typically commodity- or program-specific — and not yet appropriate for a 528 centralized, enterprise solution. Enterprise, cross-commodity applications intended for centralized use 529 across the estimating community are developed under Task 3. 530

• Design and develop custom applications and tools to support cost, schedule and 532 technical estimating/ analysis. This could include, for example, development of cost 533 and schedule toolsets/applications that could then be leveraged by analysts on specific 534 program’s estimates. 535

• Develop user-friendly interfaces that enable stakeholders to easily access and utilize 536 the developed tools and applications. 537

• In addition to project-specific analytical tools, the contractor shall develop workflow-538 integrated applications that improve cost estimator productivity by connecting CADOM data 539 products directly to estimating processes. Applications under this task are developed as 540 bounded projects, each defined and prioritized by AFCAA through the POA&M,and shall 541 leverage the data pipelines and curated data layers delivered under Task 3 rather than 542 duplicating that infrastructure. This may include: automated data pull and normalization 543 routines for recurring estimate inputs, templated analysis workflows that consume live 544 pipeline data, query and retrieval interfaces scoped to a defined estimating use case, and other 545 tools that reduce the manual effort required to move from raw data availability to analytical 546 output. The boundary with Task 3 is the boundary between infrastructure and use case: 547 enterprise capabilities that serve the estimator community broadly — such as the curated 548

CSDR data layer and general self-service access to it — are Task 3 activities; applications 549 built for a specific use case, commodity, or estimating workflow — such…