I'm Mihir, finishing my MS in Computer Engineering at NYU (May '26), looking for backend, platform, data infra, or MLOps roles. A few things I've built that I'm proud of: At McKinsey's MIO Partners, the trade-ops team was pulling data through 40+ ad-hoc scripts and waiting 14 minutes a cycle. I replaced it with a serverless data lake on Glue, Iceberg, and Athena — got that down to 6 minutes, dropped p95 fetch latency from 110s to 33s, and gave the analysts back about 120 hours a quarter. At Cimpress, I owned the pricing service behind real-time quotes across a global catalog — scaled it past 150K req/min, cut latency 30%, and rebuilt the deploy pipeline on EKS with blue-green rollouts so we went from 22-minute deploys to 13. At NYU, I've gone deeper on ML systems — RAG pipelines and a real-time voice agent running a LoRA-tuned Llama-3, where the whole game is keeping inference fast enough to feel instant. What I'm actually good at: backend that holds up under load (Node, Python, TypeScript), data infra (Spark, Iceberg, Athena, Parquet), cloud-native deployment (AWS, EKS, K8s, ArgoCD), and the unglamorous reliability work — the caching, the partitioning, the alerts that catch a problem before your users do.
