Akash Sharma

Akash Sharma is co-founder and CEO of Vellum, an enterprise platform for building, evaluating, and deploying LLM applications. After McKinsey and Dover, he started Vellum with Sidd Seethepalli and Noa Flaherty when the three kept rebuilding the same internal tooling to ship language-model features.

McKinsey, Dover, and the LLM ops gap

Sharma spent five years in McKinsey's Silicon Valley office, then joined Dover (YC S19), where the team used early LLMs in production recruiting workflows. With MIT engineers Seethepalli and Flaherty he saw that traditional MLOps did not cover prompt versioning, evals, or safe iteration once a chatbot was live. Their YC Winter 2023 Launch YC posts describe the pain as hours lost to spreadsheets and custom glue instead of product work.

Founding Vellum

Vellum started as a developer platform for prompt engineering, deployments, and monitoring, then expanded into workflows and agents for cross-functional teams. In the July 2025 Series A post, Sharma wrote that "it's one thing to build a demo and another thing entirely to get it production-ready," and that Vellum exists to close that gap. Leaders Fund led the $20M round.

On standards for production AI

He argues AI development needs a standard operating layer the way earlier generations of software got CI and observability defaults. Public talks lean on "sell before building" lessons from Dover days and on letting non-engineers participate in prompt and workflow iteration without breaking production.