Building scalable systems across backend, infrastructure, and automation
Using Contextual Multi-armed Bandits to choose the destination cluster for serverless invocations and updates choices online from observed outcomes and live metrics.
DQN agent that learns FaaS scheduling decisions from system-state signals to optimize energy–latency trade-offs (energy 7% and tail latency by 12%).
Kernel integration + benchmarking to validate speedups and correctness on real workloads.
I’m a software engineer and researcher interested in building reliable, scalable systems end-to-end backend services, cloud infrastructure, and automation. I enjoy turning ideas into clean implementations, instrumenting them with the right metrics, and iterating quickly from prototype to something production-ready.