Computer Science student at Georgia Tech with Agentic AI, network automation, C++ backend, and full-stack engineering experience at Cisco and SAIC.
Work on replayable environments for evaluating and training forecasting agents over time.
A replayable environment that pairs 1,568 resolved prediction-market events with 18.8 million dated news articles, enabling controlled evaluation and training of LLM forecasting agents.
Encrypted messaging, AI-content detection, local agents, and stock-ranking research.
An encrypted messaging product with per-message X25519 key exchange, AES-256-GCM sealed envelopes, a password-derived key vault, and safety-number verification.
A multimodal AI-content detector spanning text, image, and audio. I built the leakage-safe TF-IDF and LinearSVC text pipeline for the five-person team.
A privacy-first Windows diagnostic agent that pairs a local Gemma model with a bounded ReAct loop, typed read-only tools, and risk-labelled command cards.
A research-grade, long-only stock ranking pipeline built around point-in-time universe selection, walk-forward validation, transaction costs, and fail-closed data checks.
Software engineering at Cisco and SAIC. Machine learning research at Georgia Tech.
Built and shipped a local AI agent for network diagnostics in air-gapped environments, adopted by 200+ Cisco engineers. Combined structured diagnostic tools, command retrieval, and a fine-tuned Gemma model.
Researching economic forecasting with TinyTimeMixer and benchmarking time-series models across 120+ FRED-MD macroeconomic series.
Modernized C++ backend communications to TCP/IP and added 50+ Google Test cases, raising coverage from 15% to 35% on critical defense software.
My work spans local AI agents, C++ systems, and web applications.
Exploring how small models can help diagnose real systems with clear limits on the tools they can use.
B.S. Computer Science
Expected May 2027