
Software Engineer
Lenovo · Jul 2023 – Present
- Owned and scaled a distributed telemetry ingestion platform supporting 30,000+ devices and processing 120K+ events/min (170M+ events/day), improving throughput and storage efficiency through batching, compression, and schema optimization.
- Led the migration of a production telemetry backend from Python to Go to address concurrency bottlenecks and GIL limitations, reducing memory consumption by 93.6% (546 MB to 34.9 MB) in a 5,000-device test while maintaining backward compatibility.
- Architected a fault-tolerant event-driven architecture with RabbitMQ, implementing guaranteed message processing, retry policies with exponential backoff, and failure recovery to ensure reliable large-scale data processing.
- Designed graceful shutdown for 10+ containerized microservices, ensuring safe request draining and completion of 100% of in-flight requests during deployments, eliminating data loss and enabling seamless rolling updates.
- Led the implementation of end-to-end observability using distributed tracing (OpenTelemetry, Tempo) and centralized logging (OpenSearch), enabling 2x faster root cause analysis and reducing mean time to resolution for production incidents.
- Optimized MongoDB query performance by analyzing explain plans, eliminating full collection scans (COLLSCAN), and adding targeted indexes, reducing getCount latency from 70ms to 15ms and aggregation latency from 68ms to 26ms across 5,000-device collections.
- Optimized asynchronous task execution by tuning Celery worker concurrency and scheduling, reducing infrastructure memory usage by 70%; mentored junior engineers and conducted code reviews across the telemetry team.



