Sources Sought: Commercial Space Domain Awareness (SDA), Training Large Language Model (LLM) Module

PURPOSE/DESCRIPTION

1.  THIS IS A SOURCES SOUGHT ANNOUNCEMENT ONLY. This notice does not constitute a commitment by the Government. All information submitted in response to this announcement is voluntary, and the Government will not pay for information requested nor will it compensate any respondent for any cost incurred in developing information provided to the Government.  The Government is under no obligation to acknowledge receipt of submissions or of the information received, or to provide feedback to respondents with respect to any information submitted under this announcement.

2. This Sources Sought seeks information on commercially available large language model (LLM) trainers capable of ingesting user inputs and technical documentation to create and update training as the needs of the operational system changes. The goal is to identify commercial vendors offering LLM based trainers, today or in the future. This Sources Sought Announcement is issued solely for informational and planning purposes only and is not a solicitation.

In your response, please answer the following questions:

1. Company Information and Capabilities:
   1. Can you provide relevant experience and past performance with LLM training and development, or extensive experience deploying artificial intelligence/machine learning (AI/ML) solutions within a secure enclave, within at least IL5 FedRAMP environment?
      1. Describe your experience with fine-tuning and deploying pre-trained foundation models (e.g., Llama, GPT, Claude) for specific enterprise tasks.
      2. If you develop proprietary LLMs, describe the model architecture and the resources required to train it from scratch.
   2. Provide specific, quantifiable examples of how your deployed solutions have reduced training time or improved proficiency, preferably with metrics (e.g., 'reduced new-hire training from 4 weeks to 5 days'). Please provide specific examples of how your training frameworks have been deployed in other Department of Defense/War (DoD/DoW) systems or programs. Please provide examples of past performance that reduced training bottlenecks through on-demand instructional modules.
   3. Please provide examples of past-performance containerizing training services for classified enclaves, with full traceability to source documents and configuration-controlled updates.
2. Technical Solutions:
   1. Describe your end-to-end process for ingesting technical documents (e.g., PDFs, DOCX), TTPs, and other source materials. How does your solution assist a Subject Matter Expert (SME) in validating, refining, and approving the AI-generated training content? Can you provide on-demand job aids, updated training manuals, and assessment workflows that allow users to update training as the operational system changes.
   2. Can you generate and create curricula, learning modules, and as-needed coaching directly from doctrine and task workflows?
   3. Can you demonstrate the ability to deploy virtual instructor technologies that provide real-time guidance and feedback within classified training environments, including IL5 enclaves?
   4. How does your training solution address scalability, interoperability, and integration?
   5. How much data is required for your LLM to considered 80% trained, to provide effective solutions for training and/or modeling and simulation?
3. Compliance and Standards:
   1. Describe your methodology for ensuring AI-generated content adheres to specific doctrinal standards or regulations that will be provided. How is content traced back to source material?
   2. What specific DoD/DoW security and data handling standards (e.g., Defense Information Systems Agency (DISA) Security Technical Implementation Guides (STIGs), Controlled Unclassified Information (CUI) handling) have you implemented in past projects? Provide contract numbers or program names as examples.
   3. Is there any issue utilizing your LLM solution on a system that is fielded OCONUS (Outside the Continental United States) and will be operated by foreign nationals?
   4. Is your organization currently Cybersecurity Maturity Model Certification (CMMC) Level 2 certified?
   5. Is your organization currently Cybersecurity Maturity Model Integration (CMMI) maturity level 5 for development and services?
   6. Does your solution contain any technology controlled under the International Traffic in Arms Regulations (ITAR) or Export Administration Regulations (EAR)? If so, please describe export control considerations for OCONUS deployment and foreign national access.
   7. What is the maximum classification level your solution is currently authorized to support (Unclassified, CUI, Secret, Top Secret, Special Access Program)?
   8. Describe your organization's experience with the Authority to Operate (ATO) process for classified systems, including any existing ATOs and the authorizing agencies.
4. Innovation and Differentiation:
   1. Beyond standard Q&A, what innovative features does your solution offer to accelerate learning and knowledge retention? (e.g., simulation generation, real-time performance feedback, adaptive learning paths).
   2. How does your LLM leverage emerging technologies to enhance performance and capabilities?
   3. What future capabilities are on your development roadmap that could be relevant to our mission of on-demand training?
5. Cost and Schedule:
   1. Can you provide a rough order of magnitude (ROM) cost estimate for your proposed solution?
   2. What is your estimated timeline for development, testing, and deployment of the solution?
   3. Identify any existing Government contract vehicles (General Services Administration (GSA) Schedules, Government Wide Acquisition Contracts (GWACs), DoD/DoW Enterprise Software Initiative (ESI), etc.) under which your solution could be procured. Include contract number(s), Special Item Number(s), expiration dates, and confirmation of scope applicability.
   4. Also describe any current or recent DoD/DoW/Federal contracts where you are performing similar AI/ML training work.
   5. What is the typical development lifecycle for similar LLM training solutions from requirements definition through full operational capability?
   6. What are your industry-standard delivery schedules for initial operating capability and full operational capability?
   7. What contract type (Firm-Fixed-Price, Time-and-Materials, Cost-Plus-Fixed-Fee, hybrid, etc.) is most appropriate for your solution and why? Please explain your rationale.
6. Risk Management:
   1. What are the potential risks associated with your proposed solution, and how do you plan to mitigate them?
7. Support and Maintenance:
   1. How do you ensure long-term sustainability and reliability of your LLM solution?
   2. How stable is the connection to and through your solution? E.g., how often would you need downtime for updates and maintenance?
   3. How many support staff members are required to sustain and maintain your solution?
8. Data Rights and Intellectual Property:
   1. What are your policies and approaches regarding data rights and intellectual property for the proposed solution?
   2. How do you ensure that DoD/DoW has the necessary rights to use, modify, and maintain the solution?
   3. Can you provide Government Purpose Rights or Unlimited Rights to the training content generated by your LLM solution? Please specify which rights apply to generated content versus underlying software/models.
   4. What data rights would the Government receive for the underlying LLM models, algorithms, source code, and training datasets?
   5. Are there any restrictions on the Government's ability to create derivative works or modifications of the generated training materials? If so, please describe.
   6. What is your policy regarding Government data used to fine-tune or train your LLM models? Does your company claim any ownership rights to derivative models, weights, or parameters created from Government-provided data? Describe your approach to ensuring the Government retains full rights to models trained on Government data.
9. Collaboration and Teaming:
   1. Are you open to collaborating with other vendors or subcontractors to deliver the solution?
   2. What is your approach to managing partnerships and ensuring seamless integration of components from different sources?
10. Commercial Terms and Conditions
    1. Is your proposed solution a commercial item as defined in FAR 2.101? Please explain your rationale and provide supporting documentation if available.
    2. What are your standard commercial terms and conditions for similar LLM training solutions? Please provide a representative sample or describe typical terms.
    3. What warranties, if any, are typically offered with your solution (e.g., performance warranties, software warranties, up-time guarantees)?
    4. What are your standard commercial payment terms and conditions?
    5. Do you offer commercial pricing models (e.g., subscription-based, per-user licensing, consumption-based, perpetual license)? Please describe all available pricing structures.

SMALL BUSINESS CONSIDERATION

1. The pertinent North American Industry Classification System (NAICS) codes for this announcement are 541512 and/or 611430.
2.   Any business, regardless of size, in this area of expertise is encouraged to provide a response to this sources sought. Responses from small and small disadvantaged businesses are highly encouraged and the Government encourages teaming/partnering.
3.   This requirement involves data that is subject to export control requirements under the Arms Export Control Act (Department of State) or the Export Administration Regulations (Department of Commerce). Interested parties desiring to do business with the U.S. Space Force (USSF) shall be registered in the DoD/DoW Central Contractor Registry with a Commercial and Government Entity (CAGE) code.

 SUBMISSION INSTRUCTIONS

1.   Responses to this sources sought must be received via this Microsoft Forms Link https://forms.osi.apps.mil/r/VDKSExuQdk no later than 30 April 2026. If an interested party would like to submit sensitive or proprietary materials in addition to the Forms response, they must be coordinated with either the primary or secondary points of contact. If the response includes any proprietary information, data, or trade secrets, please clearly mark each page containing such information with an indicator, such as "Proprietary Information" or "Company Confidential." The Capability statement should be limited to no more than 15 pages. Submissions exceeding this limit may not be reviewed in their entirety.
2.   SSC support contractors are supporting the Government on this effort. As such, information submitted in response to this sources sought may be released to individuals who work SSC support contractors. Please immediately notify the primary point of contact if your company does not consent to the release of proposal information to SSC support contractors personnel.