Bhavesh Vaviya

I'm building Python backend and GenAI capabilities at BetterWorks, shipping AI features including Goal Writing Assist, AI 1:1 Summary, and real-time LLM feedback personalization.

I architected a batch AI preprocessing system that reduced output latency by roughly 10x, migrated all customers to Meta Llama 3.1, and reduced inference latency by about 35% through token sizing and prompt optimization.

I also deliver production infrastructure end-to-end, provisioning AWS MWAA with Terraform and Open Tofu, containerizing services with Docker, and establishing observability with New Relic, Grafana, and distributed tracing. I’ve resolved memory leaks and database connection-pool exhaustion under concurrent load.

Previously at Tanla Platforms, I reduced execution time from one hour to 10 minutes through Python refactoring and MySQL optimization, and improved module performance from 200 seconds to 20 seconds using Async IO. I enjoy leading technical discussions, mentoring developers, and delivering Python-heavy backend and GenAI products.