Ankur Goyal

Who is Ankur Goyal?

‍

Ankur Goyal is the founder and CEO of Braintrust, an LLM evaluation and observability platform. He is best known for building tools that help teams measure, debug, and improve AI products in production. (braintrust.dev)

Background and career

‍

Ankur Goyal has spent much of his career working on machine learning infrastructure and AI product tooling. Public sources from Braintrust and reporting on his background note that he previously founded Impira, then later joined Figma to lead ML work there before starting Braintrust. (braintrust.dev)

At Braintrust, he has focused on the hard parts of shipping AI systems at scale, especially evals, traces, and observability. Braintrust describes its product as infrastructure for turning production traces into evals and improving quality with each release. (braintrust.dev)

Key facts about Ankur Goyal include:

1. Current role: Founder and CEO of Braintrust.
2. Previous startup: Founded Impira, which was later acquired by Figma.
3. Figma experience: Led machine learning work at Figma after the acquisition.
4. Focus area: AI evals, observability, and production reliability.
5. Company focus: Braintrust positions itself as an AI observability layer for production systems. (braintrust.dev)

Notable contributions

‍

1. Founded Braintrust: Built a platform centered on LLM evals, traces, and observability for AI teams. (braintrust.dev)
2. Advanced production eval workflows: Braintrust’s product and writing emphasize using evals to compare prompts, catch regressions, and improve releases. (braintrust.dev)
3. Popularized AI observability language: His posts frame AI observability around traces, evals, and annotation. (braintrust.dev)
4. Built Brainstore: Helped introduce Brainstore, Braintrust’s custom database for high-scale AI workloads. (braintrust.dev)
5. Shared practical lessons from the field: He has written publicly about lessons learned from real eval and observability usage across customer teams. (braintrust.dev)

Why they matter in AI today

‍

1. They focus on production reality: Ankur’s work reflects what AI builders actually need after the demo phase, not just in notebooks.
2. They connect evals to shipping: His approach treats evaluations as part of the release process, which is increasingly standard for AI teams.
3. They emphasize trace-level debugging: That matters for agents, RAG systems, and tool-using workflows where failures are multi-step.
4. They make quality measurable: The Braintrust model encourages teams to define good outputs before they ship and then track regressions over time.
5. They show how infra shapes behavior: His work is a reminder that better AI systems often come from better measurement, not just better prompts. (braintrust.dev)

Where to follow their work

‍

The best place to follow Ankur Goyal’s work is Braintrust’s site and blog, where he writes about product, evals, and observability. Braintrust also publishes launch posts and technical essays under his byline. (braintrust.dev)

He also appears in public interviews and event pages associated with Braintrust, which are useful for seeing how he thinks about the AI engineering stack. (braintrust.dev)

How PromptLayer connects with Ankur Goyal's work

‍

Ankur Goyal’s work highlights why prompt iteration, evals, and production logs matter together. That is the same workflow PromptLayer is built to support, with prompt management, evaluation, and observability tools that help teams improve AI behavior over time.

Ready to try it yourself? Sign up for PromptLayer and start managing your prompts in minutes.