Contributing to the performance and reliability of real-time LLM inference systems, with a focus on reducing latency, improving throughput, and resolving infrastructure issues across distributed environments. I work across production systems to identify bottlenecks, debug system-level slowdowns, and ensure stable, high-throughput model delivery under production load

Led backend architecture at an early-stage startup, supporting over 10,000 beta users. Delivered production features and scaled the stack using async task queues and database query tuning. Helped cut response times by ~100ms and increase throughput by ~30% during peak traffic. Wore many hats, but always chased performance and stability

Built a Kubernetes performance monitoring tool using PromQL and Node.js, surfacing real-time compute bottlenecks across clusters. Improved uptime by 3% and shaved down incident resolution time by 23%. Focused on observability and backend resilience in containerized environments

Built a backend data ingestion tool in Python to track and analyze trends in cryptocurrency markets. Handled asynchronous high-frequency scraping of 1,000+ data points per cycle. Gained an early appreciation for real-time systems and performance