Projects
Things I'm building, breaking, and learning from. Some are polished, most are experiments.
Multi-Agent Due Diligence Analyst
In ProgressFour parallel LLM agents (research, financial, risk, social media) producing confidence-scored reports via tool calling and structured outputs.
Python, LLMs, SQLite, Streamlit
View on GitHub →
Series
1 That Agent Demo Won't Survive Compliance
Jun 15, 2026 2 The Gap Is the Feature: Provenance in Production AI Agents
Jun 15, 2026 3 The Score That Makes You Stop Looking: Per-Claim Confidence in Production AI Agents
Jun 19, 2026 4 The Report That Lied by Looking Complete: Graceful Failure in Production AI Agents
Jun 23, 2026 Financial Journey Intelligence Agent
In ProgressReal-time streaming system that fires an LLM agent only on meaningful behavioral triggers (~7% of events) to generate next-best-action recommendations for wealth advisors.
Python, LLMs, Streaming, Kafka
View on GitHub →
Enterprise RAG System
In ProgressProduction-grade RAG platform with event-driven ingestion, tenant isolation, ACL-safe retrieval, and grounded LLM answers with citations.
Python, RAG, Docker, Vector Search
View on GitHub →