Four build patterns (agents, retrieval, fine-tuning, and the Model Context Protocol) applied to one domain: public benefits eligibility. A wrong answer here costs a family its food or medical coverage. That stake is why every project ships a measured evaluation.
Each runs standalone and is documented and evaluated on its own. Together they compose into one intake-to-determination pipeline a state agency could run.
Deterministic SNAP eligibility exposed as auditable MCP tools. Every determination returns a result, a rule trace, and a policy citation behind it.
Grounded Q&A over the real Michigan eligibility manual, every answer cited to its exact section, refusing when the corpus cannot ground it.
A document-intake triage layer: reads an application and its uploads, screens via the rules core, and flags what needs a human. Never auto-denies.
A small fine-tune that rewrites bureaucratic notices into plain language while a faithfulness gate proves it kept every operative fact.
The MCP rules core is the deterministic decision layer the others hand the real call to. The agent screens through it. The RAG index sits inside it as a policy-lookup tool, and the notice the agent produces is what the fine-tune makes readable.
The model reads and reasons; the legal determination is made by auditable code. Each project draws that boundary explicitly, in a named graph node or a tool call.
Every answer and determination traces to a source document or a cited rule. "The model said so" is not a basis for denying benefits.
Every repo ships a hand-built labeled set and reports real numbers, including how often it fails and how it recovers.
Every project ships a synthetic generator. Handling applicant data correctly is a core public-sector competency.
A design system in the USWDS lineage, elevated. It is the shared visual language across these tools.