Software Engineer

- Designed and operated a latency-critical distributed system handling ~3500 QPS per node with sub-15ms p99, across a multi-node high-availability fleet. - Re-architected a core indexing subsystem, replacing legacy data structures with a more efficient model, delivering ~30% latency reduction validated through automated live-traffic replay and diff testing. - Led a large-scale, zero-downtime infrastructure migration across the entire production fleet OS, container images, packages, and kernel tuning. - Built an end-to-end ML model serving pipeline automated model fetching (MLFlow), ONNX Runtime inference, pre-deployment validation, and alerting; enabling faster model iteration for improved prediction accuracy and business outcomes. - Owned projects end-to-end from implementation to staged rollout, monitoring, and on-call with a consistent focus on observability, cost efficiency, and performance optimisation.