Education AI. Evidence-based, FERPA-safe.

Why an AI/ML small business for ED

The U.S. Department of Education operates at an unusual scale for a federal civilian department: it administers the Federal Student Aid portfolio (roughly $1.6 trillion outstanding across Direct Loans, FFEL, and Perkins), runs the national education statistics apparatus, funds a substantial portion of U.S. education research, and is a principal federal partner for state and local education agencies. That combination produces an AI/ML demand surface that covers everything from personalized learning pilots to loan fraud detection to national assessment analytics.

Precision Delivery Federal LLC (UEI Y2JVCZXT9HP5, CAGE 1AYQ0, NAICS 541512) is positioned as a SAM-registered small business specializing in AI and data work — and as an organization founded by a former college instructor (400+ students across 15 classes in 2025), we carry genuine domain insight into how education AI meets actual teaching, not just procurement language.

The ED components we target

• Institute of Education Sciences (IES) — the statistics, research, and evaluation arm. Hosts NCES (stats), NCER (research), NCEE (evaluation), and NCSER (special education). IES is the most research-intensive part of ED and runs a dedicated SBIR program.
• Federal Student Aid (FSA) — administers Direct Loans, Pell Grants, FAFSA, servicer oversight, and the National Student Loan Data System. Massive consumer data footprint.
• Office of Elementary and Secondary Education (OESE) — Title I administration, EDFacts coordination.
• Office of Special Education and Rehabilitative Services (OSERS) — IDEA administration, OSEP data systems.
• Office of Career, Technical, and Adult Education (OCTAE) — workforce and apprenticeship coordination.
• Office of Educational Technology (OET) — federal guidance on AI, technology, and digital equity.

Adaptive learning AI — done the evidence-based way

Adaptive learning is an established IES research focus: systems that personalize content, pacing, and assessment to individual learners. The federal bar is different from commercial EdTech: federal programs demand an evidence base — randomized or quasi-experimental evaluation, bias audits, transparent model behavior, and measurable learner outcomes.

We build adaptive systems to that standard:

• Model cards — for every predictor (item response model, knowledge tracing, recommendation).
• Bias audits — subgroup performance analysis across demographic axes required by ED guidance.
• Evaluation harnesses — A/B designs, pre-post analysis, aligned to What Works Clearinghouse evidence standards where relevant.
• Teacher-in-the-loop — adaptive systems whose outputs are transparent to teachers, not black boxes over the top of classrooms.

FSA data analytics and fraud detection

Federal Student Aid is one of the largest consumer financial operations in the federal government. Data volumes are enormous, PII sensitivity is high, and the program surface — from FAFSA application fraud to servicer compliance to Public Service Loan Forgiveness determinations — is rich with ML opportunity. Our fit:

• FAFSA and application fraud detection — anomaly detection on application patterns, identity verification signals, synthetic identity detection.
• Servicer oversight analytics — analyzing servicer performance data, call logs, complaint patterns.
• Portfolio risk analytics — delinquency and default ML on de-identified portfolio data.
• PSLF and IDR operations — LLM-assisted eligibility determination, document triage, operator tooling.

IES SBIR and research data

IES runs a distinctive SBIR program that favors education technology with measurable student outcomes. Topics tend toward classroom-ready tools, assessment systems, educator-facing AI, and data infrastructure for research. With SBIR reauthorized through September 30, 2031, IES SBIR is a stable channel. Beyond SBIR, IES operates significant restricted-use datasets (NAEP, ECLS, HSLS, NPSAS) where the research data engineering workload is substantial.

NCES and EDFacts data infrastructure

NCES administers federal education statistics programs — IPEDS for postsecondary, CCD for K-12, NAEP for national assessment — and EDFacts coordinates state reporting under federal programs. The data engineering workload is real:

• Ingesting state submissions with schema variance.
• Harmonization and edit checks at scale.
• Public-facing tools like College Navigator and state profiles.
• Restricted-use data access environments for researchers.

FERPA, PPRA, and privacy-by-design

Every ED-sector architecture we design is explicitly built for FERPA and PPRA compliance. That means data minimization, role-based access with justification, audit trails that pass OIG review, clear consent pathways for direct-to-student systems, and strict separation of identifiable vs. de-identified data environments. For IES restricted-use data, we follow the Education Sciences Reform Act and the restricted-use data license process.

Capabilities mapped to ED priorities

• Machine Learning — item response theory, knowledge tracing, recommendation systems, fraud detection — evaluation-first.
• Data Engineering — lakehouse architectures for EDFacts, FSA portfolio data, NCES. Governance and lineage.
• Agentic AI — LLM tooling for educator support, policy analysis, and operator productivity.
• Cloud Infrastructure — FedRAMP-aligned, FISMA-ready, PII-segmented.
• Cybersecurity and DevSecOps — 800-53 Moderate/High control implementation.

Past performance and honest positioning

Our confirmed federal past performance is SAMHSA (HHS) — production ML with full ATO, plus federal health IT lakehouse work. The engineering discipline translates: PII handling, 800-53 controls, governance, documentation. For ED specifically, we are targeting and pursuing work through IES SBIR, FSA modernization opportunities, and subcontracting to ED primes.

Vehicles and NAICS

• Primary NAICS 541512. Adjacent: 541511, 541519, 541690, 611710 (Educational Support Services), 541720 (Research and Development in Social Sciences and Humanities).
• Vehicles — IES SBIR, GSA MAS task orders, ED IDIQs (EDUCATE and successor vehicles), subcontracting to ED primes.

If you are an ED program office or a prime looking for an AI/ML-specialized small business subcontractor with genuine education domain experience, email [email protected].