•Developed backend for Joule For Consultants (J4C) serving 30k MAUs and orchestrating over 1M monthly messages & API calls.
•Architected granular Access Control for RAG-based vector databases, ensuring secure and precise document chunk retrieval.
•Bootstrapped Python micro-services from 0 to 1, including a multi-conversation group service and an AI-native LLM benchmarking service with unified metering.
•Optimized API latency by 60% via distributed Redis caching, significantly reducing downstream service overhead.
•Implemented Asynchronous Programming and distributed cronjobs to eliminate race conditions and maintain high data integrity.
•Spearheaded Claude Code adoption for agentic development, establishing rigorous PR review frameworks for AI-generated code.
•Streamlined DevOps by implementing Dead Letter Queues (DLQ) with real-time Slack notifications for event-driven reliability.
•Optimized CD pipelines by implementing parallel build stages to reduce deployment lead times.