Nidhi Thakkar

I’ve spent the last 8+ years leading production Python services end-to-end, from REST API design to PostgreSQL data models and message-queue-driven microservices. At Kwalee, I built highly available backend services on AWS and helped keep them reliable under high request volume.

I’ve also driven major infrastructure migrations, including moving analytics from Redshift to a serverless AWS stack (Kinesis Firehose, S3, Glue, Athena). That work processed 1M+ events/hour and reduced costs by ~70%, and I also led 2TB+ of legacy data migration into AWS cloud storage.

On the AI side, I’ve integrated LLMs, RAG pipelines, and vector search into production-style architectures. I’m building systems like ChiefOS for LLM-based email urgency classification with Gmail/Calendar OAuth2 and real-time SSE streaming, and CosmosIQ for multi-model orchestration backed by pgvector retrieval.

I’ve led and mentored teams of 8–10 engineers, running code reviews and shaping architecture decisions. I enjoy working across backend, infrastructure, and applied AI—turning system design and reliability work into features teams can ship.