Prathamesh Lohakare

I’m a product engineer at an early-stage AI-SRE startup, owning agentic developer tooling and cloud-native platform infrastructure end-to-end—from customer requirements and design through production and incident response. My north star is reliability, with a bias toward “fail loud” systems that prevent silent failure modes.

I led 0-to-1 delivery of an agentic system where an LLM writes, runs, and self-corrects production infra automation, using a peer-agent architecture that lets non-engineers run Bash/Python tasks thousands of times per day. I improved performance and correctness by cutting landing-page API latency >56% and root-causing bugs that silently corrupted session ownership records.

I use RAG to ground code generation in trusted docs and an LLM classifier to gate write-access by task risk, preventing unsafe changes. I also hardened state migration from Kubernetes CRDs to PostgreSQL, fixed a production connection-pool exhaustion incident, and closed a security leak across 7 API endpoints.

Previously, I owned platform services at 99.8% SLA for backup and metadata operations, led a monolith-to-microservices migration (REST + event-driven SQS/SNS), and built Druva’s first Public API with granular authorization for external integrations. I’ve also contributed to open source including Uvicorn, Meson, and HoloViz Panel.