§ Précis
I work the unsexy parts of LLM products — retrieval, evals, observability, cost routing — and ship them to production with measurable wins. Currently at AriesView; M.S. Machine Learning at Stevens, class of '26.
Resume ↓
§ 01 · About
Most teams know how to call an API. The hard parts are retrieval that doesn't lie, evals that don't drift, and bills that don't grow exponentially.
That's where I work. Production RAG with hybrid + graph retrieval, agentic systems wired up via MCP, semantic routers that cut cloud spend by orders of magnitude, and CI/CD gates that block hallucinations before they ship.
Open-source Model Context Protocol server. Exposes enterprise APIs to LLM agents with secure tool execution: OAuth2, token-bucket rate limiting, 500 req/min sustained. One-command Docker deployment dropped agent integration setup time by 70%.
Real-time prompt complexity classifier (<15ms) routing simple queries to local Llama 3.2 (vLLM) and complex tasks to GPT-4o. FinOps dashboard tracks per-query token cost and routing decisions. Cut cloud inference spend by 65%.
LLM-as-a-judge CI/CD gate using DeepEval — scores Faithfulness, Answer Relevance, Contextual Precision. Auto-blocks updates scoring below 95% on anti-hallucination benchmarks. LangSmith tracing across multi-agent chains; debugging time down 40%.
Autonomous multi-agent equity analyst ingesting SEC 10-K/10-Q filings, real-time prices, and earnings transcripts. LangGraph orchestration with sub-agents for DCF valuation, FinBERT sentiment, and moat assessment. Neo4j knowledge graph for cross-company comparative analysis. Generates cited Buy/Hold/Sell memos with confidence scores in under 30 seconds — cutting manual research time by 80%.
§ Education
§ Certifications