Algorithmic Trading Platform | Sub-10ms Rust Execution
Sobhan Bahrami
Algorithmic Trading Platform — High-Throughput Execution
Built the company's in-house banking and investment management platform: a time-series engine trained on 20 years of historical market data, a sentiment-analysis ingestion pipeline, and a Rust execution layer with sub-10ms order submission.
What was built
Time-series data engine: 20 years of historical data ingested, normalized, and served for model training.
Sentiment & signal models: news/social sentiment analysis feeding model training and strategy signals.
Websocket mediator: real-time market data fan-out to strategy workers.
Rust bot management: execution bots with sub-10ms submission latency and strict risk controls.
Outcome
A fully in-house trading stack — data, models, and execution — replacing fragmented third-party tooling and giving the firm complete control over latency and risk.
Stack
Rust, Python, WebSockets, time-series stores, PyTorch.
Like this project
Posted Aug 28, 2026
In-house trading infrastructure: 20-year historical data engine, sentiment models, websocket mediation, and Rust execution under 10ms.
Likes
0
Views
0
