Software Engineer - RL Gyms
Turing
Palo Alto, CA, USA (Remote)
Nov 2025 – Present
Build custom reinforcement learning training environments for enterprise clients, combining FastAPI backends with React frontiers to accelerate RL model development cycles.
- Architected RESTful API endpoints for gym orchestration - environment initialization, step execution, state snapshots, and reward tracking - handling 10K+ requests/min across distributed training clusters.
- Developed MCP (Model Context Protocol) tool integrations enabling LLM agents to interact with RL environments programmatically, opening up agentic training workflows for clients.
- Built React dashboards that surface real-time training metrics, episode replays, and hyperparameter tuning controls, cutting debugging cycles by 40%.
- Engineered FastAPI services with async workers and connection pooling, achieving sub-10ms latency on gym step operations while maintaining 99.9% uptime.