Notes that make the research program legible.
A strong AI researcher portfolio needs more than a list of publications. This page is the planned surface for technical essays, research notes, and design memos that explain how the artifacts fit together.
Formalized research memory for mathematical agents
Why a domain like coding theory needs formal statements, proof trajectories, graph memory, and MCP-style access before a theorem-proving agent can be useful.
Benchmarks as auditable processes, not files
A public-facing explanation of the EntropyMath / EntropyMaLean direction: lineage, validation contracts, solver traces, and what Lean does and does not guarantee.
From RAG to agent-readable knowledge substrates
What SOGAMBOT, MindBuddhi, regulatory RAG, and TaxCanvas-style retrieval taught me about source-grounded assistants beyond generic chatbot UX.
First goal: publish three source-backed research notes.
The notes should be short enough to read, concrete enough to cite, and connected to public artifacts rather than generic AI commentary.