Attachment 0002 - DRAID BOA PWS 30 March 2021.pdf

Joint Artificial Intelligence Center (JAIC)

Data Readiness for Artificial Intelligence Development (DRAID)

Services

Basic Ordering Agreement (BOA)

Performance Work Statement (PWS)

30 March 2021

Table of Contents

1.0 Introduction

2.0 Description of Services/Work Product Overview

3.0 General Information:

4.0 Tasks

5.0 Post Award

6.0 Definitions/Acronyms

7.0 Attachments

1.0 Introduction

1.1 Background

The Department of Defense (DoD) Joint Artificial Intelligence Center (JAIC) was established in 2018 to be the focal point of the DoD Artificial Intelligence (AI) Strategy to accelerate scaling AI and its impact across the Department. Working closely with the Services, Combatant Commands, and other Components, JAIC helps identify appropriate use cases for AI across DoD, develops capabilities, and scales impact across our enterprise with the goal of transforming the DoD through AI. The mission of the JAIC, as specified in the DoD AI Strategy, is “to accelerate the delivery of AI-enabled capabilities, scale the Department-wide impact of AI, and synchronize DoD AI activities to expand Joint Force advantages.” JAIC, the DoD, and this action also execute and operate in accordance with the AI Ethical Principles.

The purpose of this Performance Work Statement (PWS) is to help the DoD and Government users prepare data for use in AI applications by providing an easily accessible path to access the cutting-edge commercial services needed to meet the complex technical challenges involved in preparing data for AI. Through access to AI data preparation tools, capabilities, and services, the DoD will be positioned to effectively prepare AI data to support the full range of AI activities across the DoD.

The JAIC is committed to lowering the barriers to entry to this potential opportunity for non-traditional companies, especially those with limited or no prior experience working with the Federal Government or DoD. To support this, the JAIC is releasing a set of guides on executing common steps required to work with the DoD in this capacity.

Companies are encouraged to utilize these guides in responding to the JAIC DRAID

Basic Ordering Agreement (BOA) and Task Orders if needed.

1.2 Scope

The scope of this PWS shall encompass all tasks needed to create, acquire, curate, prepare, manage, or secure data sets for use in DoD AI models and application development, testing, certification, and operation. Examples of such tasks include but are not exclusive to Extract Transform Load (ETL) and Data Engineering Development, Database Design and Development, Data Science Development, Data Analysis, Data collection and curation, and Project and Outreach Program Management. For the execution of DoD Task Orders, these activities shall be performed while maintaining appropriate security and ethical considerations, including Federal and DoD regulations governing cyber security, information security, operational security, ethical standards of conduct, conflicts of interest, and AI Ethical Principles.

Capabilities and services to be procured under this PWS include:

Project and Program Management

Data Science

Data Engineering

Data Architecture

Data Acquisition and Curation

Data Quality and Analysis

Synthetic Data Generation and Data Anonymization

Software Development, Modification, and Configuration

Enterprise Information Management and Governance

Cloud Integration and Alignment

1.3. Integration of AI Data, Services, and Tools with Cloud Platforms:

The following applies to DoD components utilizing this BOA: DoD AI data services acquired or developed under this PWS and data processed by those services will frequently be required to integrate and interoperate with new or existing AI cloud platforms. In almost all cases, AI data acquired, curated, and prepared under this BOA will be required to align with published Government standards and practices for data and metadata tagging and semantics. In most cases, AI data produced under this BOA will be required to be made available on cloud platforms for reuse in other AI projects.

In many cases, AI data tools and services developed under this BOA will be required to align with and conform to published Government technical and security standards and support new or existing AI DevSecOps workflows, and in some cases may be required to be integrated into new or existing AI solutions and data pipelines. DoD components and other users are encouraged to utilize the Intellectual Property (IP) language specified in the ordering guide at the task level order.

The types of activities anticipated for aligning and/or integrating AI data and AI data tools and services include, but are not limited to, the following:

Curating and preparing AI data in accordance with DoD or Government specified

AI data and metadata standards.

Securing and encrypting AI data at rest in accordance with DoD or Government specified security mechanisms and standards.

Securing AI data tools and services in accordance with DoD or Government specified security standards to ensure they meet necessary Impact Level security standards for integration into AI Cloud Platforms.

Packaging and delivering AI data tools and services as hardened containers in accordance with DoD or Government specified standards.

Packaging AI data tools and services in accordance with DoD or Government defined standards to ensure that they can be successfully integrated into Cloud

Platforms and Architectures.

Providing AI data tools and services with DoD or Government specified

Application Programming Interfaces (APIs) for supporting data discovery, search, access, and sharing across Cloud Platforms.

Conducting data curation and data preparation support activities and providing access to curated data sets on new or existing DoD or Government platforms in support of third party AI data science and AI development activities.

To the furthest extent possible while working through and within compliance with existing Data Governance and Security procedures, data processed, and AI models trained by DoD task orders using this ordering mechanism should be made shareable to the JAIC, and the DoD at-large for reuse in future AI applications.

1.4 Type of Contract

This is a BOA. Type of contract(s) allowed under this agreement are Firm Fixed Price

(FFP) (FAR 16.202), Time and Materials (T&M) (FAR 16.601), and FFP – T&M hybrid

(FAR 16.202 and FAR 16.601). The type of contract(s) will be established by the

Ordering Contracting Officer (OCO) at the individual task order level. This BOA neither states nor implies any agreement by the Government to place future contracts or orders with the contractor or be used in any manner to restrict competition. (FAR 16.703).

1.5 Authorized BOA Users

The BOA is open for ordering to the U.S. Government in support of Artificial Intelligence

Data Readiness. This vehicle is established for decentralized ordering. The OCO is responsible for reviewing the Ordering Guide in its entirety and following all pertinent rules and regulations for awarding a BOA task order against this BOA.

1.6 Lengths of Agreements

The Data Readiness BOA will be active for five years from date of BOA execution. The

BOA will be reviewed annually, and, if necessary, updated as needed. The Government will consider off ramping contractors during this time, if it is determined that it is in the best interest of the Government. As needed, the Government will also explore on-ramping new contractors. During the final year, the BOA will be reviewed to determine if it will be renewed.

1.7 Non – Personal Services

The Government shall neither supervise Contractor employees nor control the methods by which the Contractor performs the required tasks. Under no circumstances shall the

Government assign tasks to, or prepare work scheduled for, individual Contractor employees. It shall be the responsibility of the Contractor to manage its employees and to guard against any actions that are of the nature of personal services, or give the perception of personal services. If the Contractor believes that any actions constitutes, or are perceived to constitute personal services, it shall be the Contractor’s responsibility to notify the Contracting Officer (KO).

2.0 Description of Services/Work Product Overview

2.1 Key Personnel

All DRAID BOA holders must identify a DRAID BOA Manager at the BOA level.

Qualifications for the DRAID BOA Manager are listed below:

DRAID BOA Manager: The Contractor shall identify a DRAID BOA Manager by name who shall provide management, direction, administration, quality management, and leadership of the execution of the BOA. The DRAID BOA Manager shall act as the initial interface between the OCO, Government agencies, and Contractor. The DRAID BOA Manager shall be responsible for regularly briefing the OCO and JAIC on BOA status, milestones, and provide feedback to the Government. The DRAID BOA Manager shall have the authority to respond to requirements issued through this agreement and bind the company.

It is required that the DRAID BOA Manager has the following qualifications:

Must be an employee of the prime Contractor or have an offer of employment from the prime Contractor that the individual intends to accept in the event of an award being made to the Contractor.

Demonstrate ability to lead, manage or participate as a member of a cross-functional team.

All task orders executed under this BOA must contain a Key Personnel clause for a Project Manager (PM). Additional Key Personnel may be specified as part of the task order at the discretion of the OCO.

Qualifications for the PM are listed below:

PM: The Contractor shall identify a PM by name who shall provide management, direction, administration, quality management, and leadership of the execution of the BOA. The PM shall act as the overall lead, manager, and administrator for the effort.

The PM shall direct efforts of cross-competency teams including Contractors at multiple locations, if necessary, and shall serve as the primary interface and POC with Government program authorities and representatives on technical and project issues.

The PM shall be responsible for regularly briefing leadership on program status and milestones. The PM shall oversee Contractor personnel project operations by developing procedures, planning and directing execution of the contractual, technical, multi-disciplinary engineering, programming, maintenance, and administrative support effort, and monitoring and reporting progress. The PM shall manage acquisition and employment of project resources and control financial and administrative aspects of the project. The PM shall maintain awareness of, and take responsibility for managing, the ethical and privacy implications associated with preparing or creating data sets that will be used in artificial intelligence applications.

It is required that the PM has the following qualifications:

Must be an employee of the prime offeror or teaming arrangement partner, or have an offer of employment that the PM intends to accept in the event of an award being made to the offeror.

Possess a minimum of five years of Project Management experience managing complex projects in an IT engineering or big data environment.

Possess a minimum of five years of experience managing complex projects utilizing agile project management methodologies.

It is desired that the PM has the following qualifications:

Possess Certified Information Privacy Professional (CIPP) or Certified

Information Privacy Manager (CIPM) qualifications

2.2 Labor Categories

See Appendix A labeled, “JAIC Data BOA Labor Categories Descriptions” for more information.

3.0 General Information:

3.1 Security Requirements

The Contractor shall safeguard all Government property, documents and controlled forms provided for Contractor use and adhere to the Government property requirements contained in this contract. At the end of each work day, all Government facilities, equipment, and materials shall be secured by a Government POC. Contractors are not allowed to secure Government facilities, equipment, and materials.

The Contractor shall establish and implement methods of ensuring that no building access instruments issued by the Government are lost, misplaced or used by unauthorized persons. Access codes shall not be shared with any person(s) outside the organization. The Contractor shall control access to all Government provided lock combinations to preclude unauthorized entry. The Contractor is not authorized to record lock combinations without written approval by the Government COR. Records with written combinations to authorized secure storage containers, secure storage rooms, or certified vaults, shall be marked and safeguarded at the highest classification level as the classified material maintained inside the approved containers.

Task orders outside of the DoD will need to determine specific security requirements at the task order level.

3.2 Other Security Requirements

A DD Form 254, Contract Security Classification Specification, applies to this BOA.

Task Orders issued under this BOA can support up to a Top Secret (SCI/SAP) level classification. However, specific classification/clearance requirements (Unclassified, Secret, Top Secret, etc.) will be established at the Task Order level. Unclassified Task Orders may not have a DD254 or require a facility clearance. When responding to a requirement, proposed personnel shall possess at least the minimum security clearance required for that task order prior to proposal submission or being issued a Common Access Card (CAC).

Any data or information to be publicly released (including press releases, website postings or other public statements), or shared or transferred outside the parties to this agreement must be reviewed and approved for appropriate release by the

DoD/Government Contracting Officer/OCO, the cognizant DoD Security Office (a.k.a., Protect and Counter), and the cognizant DoD Public Affairs Office (a.k.a. Strategic

Communications). Reviews and approvals shall include adherence to regulations such as the Privacy Act, Freedom of Information Act, and release of DoD information, such as DoD Instructions 5400.11, 5230.09, 3216.02, and 8580.02, and DoDD 5400.07 – these can be found at https://www.esd.whs.mil/DD/DoD-Issuances/ .

3.3 Government Furnished Property

Government furnished items or services will be specified by the OCO in the individual task orders.

3.4 Data Rights

Data rights shall be specified at the task order level by the ordering activity.

3.5 Service Contract Reporting (SCR)

The Office of the Secretary of Defense (OSD) mandates reporting of manpower data with regard to the performance of DoD services contracts consistent with the requirements of title 10, U.S.C., Section 235 and 2330a, as amended. Contractors are to report manpower data relating to the performance of services contracts into SAM, consistent with existing Service Contract Reporting (SCR) requirements under the FAR

Subpart 4.17 - Service Contracts Inventory (SCI). When reporting on task orders issued under an indefinite-delivery contract or agreement, reports are completed at the order level. No reports are collected at the contract or agreement level. Contractors are required to report data in SAM on an annual basis when awarded a DoD contract or task order valued in excess of $3 million in obligations. The Contractor shall completely fill in all the information in the format using the following web address: http://sam.gov.

SCI reporters are required to report the following information: Total Amount Invoiced, Prime Contractor Hours Expended, and, if applicable, Tier 1 subcontractor information.

The reporting period for Contractors in SAM is open from mid-October through mid-

December for reporting against the prior government fiscal year. Contractors shall complete the SCR before the reporting period closes against the prior government fiscal year.

3.5.1 Contracting Officer’s Representative (COR)

No COR will be established at the base BOA level. Each task order will have a separately identified COR and designation letter that specifies their specific authorities and responsibilities. The COR will be identified by the designation letter issued by the KO. A copy of the letter will be sent to the Contractor and the

Contractor will acknowledge each COR’s designation letter. Under no circumstances does a COR have the authority to authorize substantive changes to https://www.esd.whs.mil/DD/DoD-Issuances/ http://sam.gov/ any terms and conditions of the resulting order with respect to cost/price, and delivery dates.

3.5.2 Quality Control

Contractors performing under this BOA shall ensure that they have an approved

Quality Management System that is certified to standards determined at the task order level. The Contractor shall develop a Quality Control Plan (QCP) within 30 day of Notice To Proceed (NTP) of a task order under this BOA. For DoD Task

Orders, the QCP shall include evaluation methods for Responsible AI and/or adherence to the Department of Defense AI Ethical Principles adopted February

2020 (“DoD AI Ethical Principles”; see Appendix B).

3.5.3 Quality Assurance [Surveillance Plan]

All task orders under this BOA shall include a Quality Assurance Surveillance Plan

(QASP) outlining the Government’s surveillance methodology and assessment of the Contractor’s performance in terms of an acceptable level of quality. For DoD

Task Orders, the QASP shall include evaluation methods for Responsible AI and/or adherence to the DoD AI Ethical Principles.

3.5.4 Business Size Standards

Contractors must certify Business size standard for each task order. A size standard, which is usually stated in number of employees or average annual receipts, represents the largest size that a business (including its subsidiaries and affiliates) may be to remain classified as a small business for SBA and federal contracting programs. The definition of “small” varies by industry.

All BOA holders’ size standards will be confirmed by the OCO at the task order level for eligibility to compete.

3.5.5 Subcontracting Plans

The Small Business Subcontracting Plan requirement is required for Contractors with a business size designation of “other than small” in accordance with FAR

Clause 52.219-9, Small Business Subcontracting Plan, Alt II and DFARS clause

252.219-7003, Small, Disadvantaged, and Women-Owned Small Business

Subcontracting Plans (Government Contracts). Unrestricted task orders shall include Small Business Subcontracting Plan requirements. The Small Business

Subcontracting Plan shall be defined by the OCO at the task order level.

3.5.6 Small Business Participation

The Small Business Participation Plan shall be defined by the OCO at the task order level.

3.6 Recognized Holidays

Work at a Government site shall not take place on Federal holidays or weekends (but may require off-hour work due to network loading or other disruptions that could occur) unless direct by the OCO on individual task orders.

4.0 Tasks

This is not an all-encompassing list, as technology changes, additional tasks could be required.

Required Task Selections (Tasks 1 – 4)

The following tasks are required as part of every task order. The Contractor shall provide General Services for these required tasks.

4.1. Task #1: Project Management and Documentation

The contractor shall provide project management to an assortment of initiatives necessary to leverage enterprise data sets as strategic assets for the development of

Data Science and Artificial Intelligence applications. The contractor shall provide project management and oversight of initiatives including, but not limited to, data acquisition, ingest, data governance and metadata management, integration and preparation of data, including labeling, quality assessment, and when executing within

DoD Task Orders, data minimization and other practices in support of the Department of

Defense DoD AI Ethical Principles. The contractor shall also provide project management for efforts involving any selection from the additional optional tasks below.

If specified at the Task Order, the contractor shall plan and coordinate events including, but not limited to, hackathons, programming competitions, and data science outreach events that leverage these technologies. The contractor shall provide regular updates including project plans, status updates, risk assessments, Data Governance plans, and progress towards major milestones.

The contractor shall provide technical documentation for data sets handled and processed. The contractor shall author research style data quality evaluations and technical methodology papers to facilitate knowledge transfer.

The Contractor shall also provide corresponding documentation for business process owners and end users describing, but not limited to, metadata, handling, processing, use, and limitations of data sets.

4.2. Task #2: Software Configuration

The Contractor shall configure and secure software as needed.

4.3. Task #3: Cloud Integration

The Contractor shall integrate developed or delivered solutions into new or existing cloud architectures and platforms as needed.

As part of this integration, DoD task orders shall be required to complete JAIC Data

Cards and maintain compliance with relevant DoD data catalogs.

4.4. Task #4: Department of Defense AI Ethical Principles

When executing on DoD Task Orders, the Contractor shall demonstrate how its product or solution will consider, address or instantiate the DoD AI Ethical Principles and aids in mitigating ethical risks.

Optional Task Selections (Tasks 5 – 13)

The following tasks are optional as part of every task order. The Contractor shall provide General Services for all optional tasks as indicated in the task order.

4.5. Task #5: Data Ingestion, Integration, Architecture, and ETL

Subtask 5.1: Data Ingestion Services

The contractor shall provide data ingestion services to include the establishment and development of new data pipelines utilizing technologies including but not exclusive to web services and streaming data services. These data transfer pipelines shall be designed as interchangeable and reusable components that may facilitate data exchange for multiple Data Science and Artificial Intelligence projects. The contractor shall develop methods to secure the data transfer to the

Impact Level (IL) of the data and provide transfer capabilities to move data between systems of different Impact Levels. These capabilities are to include fine grained user access controls for data at rest and secure encrypted technologies for data in transfer.

Subtask 5.2: Data Architecture and Preparation

The contractor shall provide a scalable technical architecture necessary to enable the application of prepared data sets by Data Science and AI workflows. This architecture shall not result in reduced performance of data source systems below user requirements because of increased data consumption workloads.

Additionally, the architecture shall be documented in entity relation diagrams and supporting documentation to maximize usefulness of processed data sets.

The contractor shall perform all Data Preparation tasks necessary to transform data from its raw form into a format ready for consumption by Data Science and AI workflows. These tasks include, but are not limited to, data export standards and

Extract-Transform-Load activities to maximize data usage for AI and analytical usage.

The contractor shall act as an integrator of technologies and solutions. They shall provide technological solutions that enhance and complement existing products currently in use, as well as future technologies provided by outside vendors.

Subtask 5.3: Data Integration

The contractor shall develop an enterprise data integration model that defines the data flow requirements for sensitive and restricted information. The model must address all potential data flows internal and external to the enterprise. The contractor will develop or provide technologies which allow for the integration of various data sources into common visualization and data science platforms and libraries.

The contractor shall also provide integration techniques to improve the quality and usefulness of data collections. Techniques such as Master Data Management and

Data Enrichment shall be utilized to enhance data and assist increase its predictive power in Data Science and AI applications.

Subtask 5.4: Standardized Data Science and AI Pipelines

The contractor shall leverage existing infrastructure to develop and customize standardized Data Science and AI tools and pipelines which can be leveraged for multiple projects. Pipelines shall be deployed with flexible tools allowing for future scale-ability and growth. The tools should be available as micro-services in a containerized environment and should be designed to be inter-operable with new and existing technologies.

Subtask 5.5: Automated Data Collection

The contractor shall work to improve data quality by implementing applications and pipelines for the automated capture and collection of data including technologies such as, but not exclusive to, sensors, IOT, edge devices, or other intelligent data capture devices.

Subtask 5.6: Streaming Transformation

The constructed pipelines should have capabilities for streaming data transformation including but not exclusive to joining, merging, filtering, and performing Change Data Capture techniques on ingested data sets.

Subtask 5.7: Data Storage

The contractor shall design very large capacity, scalable data storage systems including secure user and role based access controls.

Subtask 5.8: Data Security

The contractor shall provide technologies for assessing the security of the data with respect to its impact on the security of the AI system. This shall include methods to detect and remove data designed to interfere with the training and/or operation of an AI model or application, as well as securing the data throughout all points of the data lifecycle.

Subtask 5.9: Metrics and Documentation

Key data ingest deliverables shall be provided by the contractor in relation to the size of data in both storage space as well as number of data entities, and the timeliness of data refresh. Data lineage and provenance shall be recorded at all steps of processing. This documentation is to include access procedures.

4.6. Task #6: Data Accessibility and API’s

The contractor shall provide technologies and resources which allow for data and storage redundancy while providing data security as indicated by the appropriate impact level. The contractor shall make data available through the development of multiple interface forms including but not exclusive to API’s, web services, Secure File Transfer

Protocol (sFTP) transfers, cloud based data buckets, edge devices, and database

ODBC/JDBC connections. These interfaces shall be designed to encourage high levels of data sharing and collaboration while maximizing the security of sensitive data sources.

4.7. Task #7: Enterprise Information and Metadata Management

Subtask 7.1: Cataloging Data

The contractor shall catalog metadata for all data ingested in the readiness effort.

Complete collections of metadata and metrics for data assets including but not exclusive to, table data and elements such as columns shall be gathered. The contractor shall also provide administrative support for the collection and documentation of data element definitions and entity relationships.

Subtask 7.2: Making Data Visible

Metadata and metrics collected shall be formatted to facilitate searching, display in, and portability between Enterprise Information Systems. These collections shall be organized in searchable interfaces such that users may be able to easily locate organizational information and its metadata. The contractor shall also provide visibility into cross domain data relationships through integrated entity relationship models that show relationships between data generated or stored in disparate systems.

Subtask 7.3: Metadata Integration

The contractor shall ensure that the collection, formatting and storage of metadata and metrics meets all existing format standards for the awarding Government

Organization and integrates with existing metadata catalog tools.

Subtask 7.4: Transfer Learning and Algorithm Management

The contractor shall manage and maintain an organized library of trained algorithms to facilitate the responsible reuse and rapid development of artificial

Intelligence solutions. The contractor shall include storage structures, maintenance, and documentation including metadata, statistics, and associated review and governance processes.

4.8. Task #8: Data Quality, Feature Engineering, and NLP

Subtask 8.1: Data Quality

The contractor shall provide technologies for assessing the quality of data metrics gathered through this assessment shall be displayed in reporting interfaces and data readiness dashboards. Data Quality may be assessed through processes including but not exclusive to, Outlier Analysis, Sparsity Analysis, and additional statistical methodologies.

Subtask 8.2: Data Wrangling

The contractor shall perform data wrangling including all activities necessary to profile, connect or join, filter, and transform data into Machine Learning and AI

Feature Sets. The contractor shall write custom data wrangling code where applicable and shall also leverage existing data processing tools and pipelines in this effort.

Subtask 8.3: Feature Engineering

The contractor shall perform development of Data Science and AI features by transforming data sets using programming languages, libraries, and ETL tools. The contractor shall process, filter, join and integrate all data sources as needed and use techniques to provide data structures that meet the input needs of Data

Science and AI algorithms. The contractor shall seek to increase the power of such algorithms by using combinations of data elements to create calculated features.

Subtask 8.4: Modality-specific Processing and Preparation Tasks

The contractor shall perform processing and preparation of data across relevant modalities, to include, but not limited to, image, full motion video, and text. For example, for language-based data modalities, the contractor shall perform processing of structured and unstructured text data to create feature sets for

Machine Learning and AI applications, and perform tasks to include, but not limited to, text reduction, extraction, tokenization, mining efforts to prepare data for AI pipelines, and building of applications to tag and label text data.

Subtask 8.5: Data Representativeness

The contractor shall provide technologies for identifying a lack of data representativeness (or alternatively, identification of bias) in testing and/or training datasets. This shall include methods to detect general data bias resulting from incidents such as insufficient data collection and storage, poor data processing, data drift, or flawed feedback loops.

4.9. Task #9: Synthetic Data

The contractor shall provide technologies for generating synthetic data within user-defined statistical constraints. The contractor shall develop methodologies for validating statistical similarity and lack of any secure or confidential original data. Generated synthetic data shall maintain originating entity relationships and statistical or distribution lineage to its source such that developed and trained algorithms may be used with interchangeability between data sets.

4.10. Task #10: Data Anonymization

The contractor shall provide technologies for anonymizing data within user-defined statistical constraints. The contractor shall provide methodologies for validating the degree to which the data has been anonymized as required by the user. This may include but is not limited to protecting personal information and protected health information, sensitive data, technical data, intellectual property, or proprietary data, as well as to meet Federal and DoD regulations. The level of anonymization will be provided at the task order level if applicable.

4.11. Task #11: Data Labelling

The contractor shall provide methods and technologies capable of providing proper labels to datasets that require labeling. The contractor shall provide methods and technologies that are capable of evaluating and reporting on the accuracy and confidence of automated labeling efforts when such technologies are used. Finally, the contractor shall provide methods and technologies of evaluating the percentage of missing, correct, and incorrect labels on any previously labeled data.

The contractor shall support labeling efforts by developing applications in support of this effort, including object detection applications and pipelines.

The contractor shall seek out and deploy technologies which assist in, automate, or reduce the labeling effort. Technologies such as, but not exclusive to, weak supervision automated or assisted labeling and data reduction algorithms should be utilized to their maximum potential to build highly efficient reusable labeling pipelines.

The contractor shall create, update, integrate, or utilize ontologies as required by the user.

4.12. Task #12: AI Security

The contractor shall provide methods and technologies capable of providing security for

AI algorithms to protect them and their functionality from deliberate manipulation or coercion intended to impair or influence outputs in an unfavorable manner.

4.13. Task #13: Data Science and AI Outreach and Innovative Development

Mechanisms

The contractor shall leverage the processed and generated data sets for outreach projects and innovative AI application development mechanisms, such as competitions that leverage development inputs from academic and industry communities. A competitive development model shall be designed that allows for outside entities and individuals to engage in hackathons and competitions using synthetic or unclassified openly available data in a manner that encourages innovative solutions. The contractor shall plan and execute these events.

5.0 Post Award

5.1 Conferences

As needed, the Government will hold post-award conferences. The Contractor shall attend any post award conference convened by the contracting activity or contract administration office in accordance with Federal Acquisition Regulation Subpart 42.5.

The contracting officer, PM, Contracting Officer's Representative (COR), and other

Government personnel, as appropriate, may meet periodically with the Contractor to review the Contractor's performance. At these meetings, the contracting officer will apprise the Contractor of how the Government views the Contractor's performance and the Contractor will apprise the Government of problems, if any, being experienced.

Appropriate action shall be taken to resolve outstanding issues. These meetings shall be at no additional cost to the Government.

5.2 Transition In/Transition Out

Transition in and out for contractors on this BOA will be specified for each task order.

5.2.1 Transition-Out Plan.

The Contractor shall provide a Transition-Out Plan NLT than 30 calendar days prior to expiration of the contract. The plan shall facilitate the accomplishment of a seamless transition from the incumbent to an incoming Contractor (if applicable).

In addition, the Contractor will continue to accomplish all tasks as outlined in the contract during this period. The current Contractor shall identify how it will coordinate with the incoming Contractor and Government personnel to transfer knowledge regarding the following:

• Project management processes

• Points of contact

• Location of technical and project management documentation

• Status of ongoing technical initiatives

• Transition of key personnel

• Schedules and milestones

• Actions required of the Government

• Coordination of IT related programs, issues

5.2.2 Communication.

The Contractor shall also establish and maintain effective communication with the

Government personnel for the period of the transition via weekly status meetings per the COR’s direction.

6.0 Definitions/Acronyms

Acronyms and Definitions shall be derived from Federal and DoD resources, such as statutes, instructions, directives, and policies. Deviations must be pre-approved by the

DoD/Government Contracting Officer/OCO.

7.0 Attachments

Appendix A. JAIC DRAID BOA Labor Categories Table

Appendix B. Department of Defense AI Ethical Principles