A muti-agent Dnd DM that directs you to dive into a fantasy world!

Structure-first pipeline that transforms GitHub repositories into question-indexed, graph-grounded semantic assets for code search, Graph-RAG, and LLM-based reasoning.

PromptGuard is a pragmatic, opinionated framework for establishing continuous integration for LLM behavior. It operates on a simple, verifiable principle: run the same prompts across multiple model configurations, compare outputs against defined expectations, and flag semantic regressions.

Research-grade neuro-symbolic RAG framework where retrieval is a policy, not a vector search, built for evaluation, ablation, and reliability analysis.

A systems-first implementation of agent control: explicit retrieval decisions, planner–executor separation, and auditable memory as core architectural mechanisms.

Production-grade Retrieval-Augmented Generation (RAG) system for evidence-grounded equity research over SEC filings and live market news.

Failure-first analysis of retrieval-augmented and agentic systems, focused on isolating and attributing failures across retrieval, planning, execution, memory, and policy layers.

Not a hype showcase. Just solid AI workflows, prompt structures, and automation systems I build, use, and improve. Hands-on. No fluff.

A conservative, agentic RAG system with evidence-bounded synthesis, citation enforcement, and zero hallucination tolerance.

Documentation and design artifacts for Channel AI (Formerly New Dhatu), a multi-agent conversational analytics system.

Automatically extracts, validates, and structures invoice data from images and PDFs using OCR + a local LLM—eliminating manual data entry.

Explicit control and observability over when an LLM should answer, hedge, or refuse — treating generation as a governed system layer, not a side effect of retrieval.

A systems-level analysis of static RAG pipelines, isolating ingestion, retrieval, and ranking boundaries to expose structural failure modes before generation.