A self-improving multi-agent quant research system. In this post, I explain how I built a human-guided agentic workflow that takes a trading idea through research, implementation, backtesting, independent validation, and, if it passes every gate, deployment to live markets.
Read moreA flexible, open-source charting tool I built for my quant research. It renders candles, trades, stops, and indicators in one HTML file, zooms like a paid trading platform, and can be driven by AI agents that add overlays, pick the window, and label what to review.
Read more Open sourceSeven checks I use before a trading idea can move from research to capital. They cover code validity, walk-forward testing, trial counts, sealed evaluation, shadow execution, and risk limits.
Read moreA 15-second labeling error changed two years of backtest results. Lag correlation exposed the shift, and a clean rebuild showed which strategies were sensitive to bar construction.
Read moreHow I turn 24 months of raw NQ data, over 300 GB of ticks and order-book updates, into compact Parquet datasets, cache shared indicators, and use NumPy and Numba to make repeated backtests fast.
Read moreA self-improving multi-agent quant research system. In this post, I explain how I built a human-guided agentic workflow that takes a trading idea through research, implementation, backtesting, independent validation, and, if it passes every gate, deployment to live markets.
Read more quant · AI engineering · 10 min Dec 5, 2025 PinnedA flexible, open-source charting tool I built for my quant research. It renders candles, trades, stops, and indicators in one HTML file, zooms like a paid trading platform, and can be driven by AI agents that add overlays, pick the window, and label what to review.
Read more quant · open source · 3 min Open source Feb 15, 2026Seven checks I use before a trading idea can move from research to capital. They cover code validity, walk-forward testing, trial counts, sealed evaluation, shadow execution, and risk limits.
Read more quant · research systems · 8 min Jan 3, 2026A 15-second labeling error changed two years of backtest results. Lag correlation exposed the shift, and a clean rebuild showed which strategies were sensitive to bar construction.
Read more quant · data engineering · 9 min Dec 12, 2025How I turn 24 months of raw NQ data, over 300 GB of ticks and order-book updates, into compact Parquet datasets, cache shared indicators, and use NumPy and Numba to make repeated backtests fast.
Read more quant · data engineering · 11 minMy research focuses on distributed algorithms for large-scale networks, distributed systems, self-adjustment, cloud computing, and IoT, evaluated both theoretically and experimentally. Ph.D. under Professor Sukumar Ghosh, The University of Iowa.
A selection of things I've built, from research prototypes to production systems.
Independent research: backtesting, alpha discovery, and real-time monitoring over ~100B tick-level datapoints, with AI agents in the research loop.
A web interface for spatial analyses on Twitter data, backed by a sharded MongoDB cluster.
GPU-accelerated kNN over 500M+ data points in 160 dimensions for pixel-level skin detection.
Python harness that auto-runs student submissions against spec; used for 150-student sections at Iowa.
Android app that reads words from subtitles in real time and stores them as spaced-repetition flashcards.
Network-science analysis of giant components, degree distributions, and diameters in research collaborations.
Globus-based grid with PBS/SGE clusters and a friendly Java portal; undergraduate thesis work.
Case-handling system for the Bangladesh Supreme Court with four user roles. Best Project, IUT ICT Fest 2008.
Courses I've instructed at the University of Iowa and Cornell College, where I served as a Visiting Assistant Professor of Computer Science.