Currently.

• 2026 WIP

Previously.

• Volatility Surface Forecasting: ConvLSTM neural network forecasting S&P 500 implied volatility surfaces, achieving 23% lower MSE than GARCH baseline
• Deep Learning for Intraday Volatility: Training GRU + HAR + GARCH models on SPY/QQQ realized variance and order book imbalance to predict intraday volatility
• ML Enhanced Risk Parity: Portfolio construction using risk parity with machine learning forecasted covariance matrices
• Low-Latency Trading Systems: High performance trading infrastructure in Rust and C++ for real-time market data processing with sub-microsecond latency
• Graph Neural Networks for Order Books: Building GNN transformers pretrained on 1M+ contact graphs for financial market prediction
• iOS Security Research: Developing Pointer Authentication Code bypass techniques for iOS jailbreak development and kernel exploitation
• NASA SUITS: Architecting local offline AI assistant for EVA spacesuits using BitNet 1.58-bit quantized models with multiagent system