Low-latency, high-throughput market data pipeline and trading system for crypto markets.

• Achieved 15+% annual returns with 2% max drawdown over 1 year.
• Processed 1k+ trades/minute with low latency.
• Handled $10M+ from 1k+ subscriptions

Event-driven crypto perpetual futures trading system with advanced backtesting and research capabilities.

• Achieved <500ms latency from signal generation to order execution
• Deployed on GCP using Docker and CI/CD with 99.5% uptime
• Implemented robust, portfolio-level risk management controls, maintaining max drawdown <2%

Podcast role identification data pipeline using prompt engineering and analysis of the Social Network of Hosts and Guests in Podcasts Ecosystem.

• Identify hosts and guests in 1M+ podcast episodes with 80% recall
• Reduced processing time by 144x using vLLM and batch processing
• Improved LLM recall by 20% and reduced processing time by 40% by advanced prompt engineering

Implemented robust REST API endpoints using Go (Gin framework) and designed PostgreSQL database schemas.

• Enabled customer-facing strategy subscription and configuration platform.
• Supported frontend applications to visualize trading strategies performance.