I build AI-powered products, from LLM orchestration and agentic backends to user-facing interfaces. My focus is shipping systems that are production-ready: low latency, observable, and actually useful.

• Currently building: advanced Agent Orchestration workflows and AI developer tooling
• Ask me about: LLM system design, agentic pipelines, RAG architecture, MLOps
• Core strengths: End-to-end product ownership · Latency optimization · Scalable AI system design

AI-Powered Codebase Visualization

• Interactive 2D force-directed graph of GitHub repo call structures, powered by the Claude API.
• Built with React, D3.js, and a custom graph traversal backend.
• Tech: React, D3.js, Claude API, Python, GitHub API

AI Resume Tailoring Pipeline

• Three-agent pipeline (Tailor, Validation, Formatting) that rewrites resumes against job descriptions with ATS scoring.
• Full-stack app with React/Express frontend, deployed at dub.sh/custom-resume.
• Tech: Claude API, React, Express, Multi-Agent orchestration, Docker

Privacy-Preserving Local LLM Email Analysis

• Hybrid architecture: local LLM (LM Studio + Gemma 3) for on-device processing, Claude API for reasoning on sanitized signals.
• MCP server query layer with SQLite storage.
• Tech: LM Studio, Gemma 3, Claude API, MCP, SQLite, Python

Natural Language AdTech Backend

• Text-to-SQL agent over unified Snowflake + MongoDB — no SQL needed for analysts. 4-6x latency reduction via multithreaded execution.
• Tech: FastAPI, Text-to-SQL, Snowflake, MongoDB, Python

Large-Scale Academic Research Analysis

• TF-IDF trend mining over 10 years of academic papers with graph modeling to surface emerging domains.
• Tech: NLP, TF-IDF, Topic Modeling, Graph Analytics, Python