Developed research infrastructure and internal tooling for an early-stage MFT pod
- Owned in-house backtest platform that allowed researchers to iterate on alphas at scale
- Parallelised backtests by provisioning dedicated VMs for each job, tightening feedback loop from hours to minutes
- Extended platform to support custom trading universes, trading restrictions and index hedging.
- Reduced cloud costs ~60% with service that restarts preempted VMs and monitors CPU/Memory utilization
- Built frontend to simplify managing and monitoring of VMs for researchers
- Created Terraform configurations for multi-dependency deployments on GCP
[Python, Polars, Redis, GCP Compute Engine/Spanner/Cloud Scheduler]
Worked on backend services supporting TikTok Shop's logistics between sellers and delivery hubs
- Enabled new market launches by implementing region-specific features and multi-region deployments
- Improved service decoupling by migrating inter-service communication from RPC to Kafka
- Reduced API response latency by migrating from polling to server-side push
- Responded to and resolved production incidents during EU hours
[Go, Kafka]
Worked on latency-sensitive features for interfacing with Autonomous Vehicles [C++]
Built from scratch execution and market data systems for Binance/OKX [Python, Kafka]
Built a proof-of-concept C++ FIX matching engine for aggregating FX swap quotes across venues [C++]
Prototyped an event-driven architecture using Apache Airflow/RabbitMQ for task scheduling [Python]
C++ aggregator streaming Polymarket BTC Up/Down market data and Binance aggTrade