Mathematics & Computing Undergraduate @ IIT Patna
Bridging the gap between Stochastic Calculus and Bare-Metal C++ Engineering.
I operate at the intersection of High-Frequency Trading (HFT) and Deep Learning. My goal is to build autonomous trading agents that can reason about market liquidity and execute trades with sub-millisecond latency.
| The Engine (Systems) | The Brain (Quant) |
|---|---|
| Low-Latency C++ | Stochastic Calculus |
| Market Data Feeds | Deep Learning (CNN/LSTM) |
| Matching Engines | Time-Series Analysis |
| Network Optimization | Market Microstructure |
| Languages | Quant Libraries | Infrastructure |
|---|---|---|
 |
 |
 |
 |
 |
 |
 |
 |
 |
A Hybrid Execution System combining C++ speed with Deep Learning intelligence.
- Architecture: Decoupled C++17 execution engine + Python PyTorch Inference Server.
- Strategy: DeepLOB (Deep Limit Order Book) utilizing CNNs for spatial features and LSTMs for temporal patterns.
- Performance: 57.9% Win Rate | 1.55 Profit Factor on Binance Futures (Live Test).
A simulation of a NASDAQ-style matching engine optimized for cache locality.
- Core: Price-Time Priority matching algorithm using
std::vectorand memory pools. - Optimization: Fixed-point arithmetic to eliminate floating-point non-determinism.
- Throughput: Capable of processing 10k+ orders/second in single-threaded backtests.