Why HHS is where our past performance is the deepest
Most AI small businesses pitching federal health work do not have a production federal ML system in their past. We do. Our founder delivered a production machine learning system at SAMHSA — through full ATO, on sensitive federal behavioral health data, serving real users, not a proof-of-concept that expired in a sandbox.
Precision Delivery Federal LLC (UEI Y2JVCZXT9HP5, CAGE 1AYQ0, SAM.gov active, SBA small business) is the corporate vehicle built around that capability. We focus on HHS AI work where the blend of production ML, data engineering, governance, and ATO-minded design is the durable differentiator. See the detailed past performance page.
HHS operating divisions we target
HHS is a $1.7T+ agency with 13 operating divisions and more than 80,000 employees. Our focus is on the divisions with the highest AI/ML and health-data-platform demand:
- SAMHSA — Substance Abuse and Mental Health Services Administration. Our past performance anchor. Behavioral health data, treatment locator, state grant data, public health surveillance.
- NIH — National Institutes of Health. 27 institutes and centers. SBIR and STTR through eRA Commons. AI/ML relevant across every biomedical domain.
- CDC — Centers for Disease Control and Prevention. Public health surveillance, outbreak detection, syndromic surveillance, data modernization (DMI).
- CMS — Centers for Medicare and Medicaid Services. The largest federal healthcare payer. Claims analytics, fraud detection, quality measurement, program integrity.
- FDA — Food and Drug Administration. AI/ML as a medical device review, pharmacovigilance, inspection prioritization.
- HRSA — Health Resources and Services Administration. Grant program data, workforce analytics.
- ACF — Administration for Children and Families. Program data, evidence synthesis, SBIR.
- ASPR — Administration for Strategic Preparedness and Response. Emergency response data, BARDA-related work.
- IHS — Indian Health Service. Health IT modernization, clinical data platforms.
HHS mission areas mapped to our capability
Public health surveillance ML
Anomaly detection on syndromic data, outbreak signal amplification, data quality ML for state-reported feeds. CDC and SAMHSA-relevant. Kaggle Top 200-level modeling, federal-grade governance.
Clinical documentation and review AI
Agentic LLM systems for clinical note review, evidence synthesis, guideline adherence, and administrative burden reduction. See Agentic AI.
Claims analytics and fraud detection
Supervised and unsupervised ML on CMS-scale claims data, network and graph approaches, rare-event detection. Large-data engineering substrate.
Behavioral health data platforms
SAMHSA-adjacent work we know directly. Governed data lakes, privacy-preserving analytics, 42 CFR Part 2 awareness.
Grant review and FOIA / inquiry triage
Agentic LLMs with RAG over federal document corpora and human-in-the-loop review gates. Evidence synthesis for policy analysis.
Health data platform modernization
Lakehouse architectures, governed analytics, real-time streaming. Our federal health IT data platform past performance transfers directly.
HIPAA, 42 CFR Part 2, and FISMA — we know this ground
HHS AI work has a compliance surface most federal-curious AI firms underestimate. Beyond FISMA and NIST 800-53, HHS data touches:
- HIPAA Privacy and Security Rules — protected health information handling, minimum necessary, access controls, audit logs.
- 42 CFR Part 2 — heightened protections for substance use disorder records. SAMHSA-central. We know this first-hand.
- FISMA Moderate and High — most HHS systems run FISMA Moderate; some FISMA High.
- SORN and Privacy Act — Systems of Records Notices for systems containing records retrievable by personal identifier.
- IRB and data use agreements — for research-adjacent work including NIH-funded scopes.
We design AI systems around these controls from day one, not bolt them on.
NIH SBIR, eRA Commons, and grant pathways
NIH runs the largest non-defense SBIR program. Key characteristics for small businesses:
- Higher Phase I and Phase II budgets than DoD on many topics.
- Rolling and standard-date submissions — less feast-or-famine than DoD DSIP windows.
- eRA Commons — the electronic Research Administration portal. Registration required for the firm and PI.
- Direct-to-Phase-II option — for firms with a demonstrated Phase I equivalent.
We are targeting NIH SBIR topics in AI/ML on clinical data, behavioral health, and public health. Research.gov (NSF) registration is already complete in parallel.
CMS, CDC, and the data modernization wave
CMS is moving aggressively on cloud modernization and AI-assisted program integrity. CDC's Data Modernization Initiative (DMI) is a multi-year rebuild of the public health data pipeline. Both are friendly environments for a small business with a production federal ML system in hand and governed data platform experience. We pursue these through NAICS 541512 small business subcontracting to primes and through direct engagements where the scope fits.
Past performance and honest positioning
For HHS specifically, we can lead with confirmed past performance, not aspiration:
- SAMHSA production ML system — shipped, ATO'd, real users, sensitive behavioral health data.
- Federal health IT data platform — lakehouse architecture, governance-first.
- Multi-agency cloud migration — delivered through prior consulting employers, including health-adjacent federal workloads.
For HHS operating divisions where we do not have direct past performance (CMS, CDC, NIH as specific examples), we are targeting and pursuing work, and we bring transferrable production ML and health-data governance from SAMHSA.
Relevant NAICS and vehicles
- Primary NAICS 541512 — Computer Systems Design Services. SBA small business.
- Adjacent — 541511, 541519, 541690, 518210, 541612 (Human Resources Consulting) for some workforce analytics scopes.
- Vehicles — NIH SBIR/STTR via eRA Commons, HHS SBIR where applicable, subcontracting on CMS and CDC IDIQs, HHS BPAs, and OTAs (ASPR and BARDA have OTA authority).
How HHS program offices and primes engage us
Three patterns:
- NIH SBIR / STTR prime or partner — on AI/ML topics where we can carry scope or pair with a research institution.
- HHS subcontracting — as an AI/ML specialty subcontractor with confirmed federal health past performance.
- Direct engagement — for program offices that want a small AI specialist with ATO-ready practices.
Email [email protected]. We respond within 24 hours.