Parlance Labs

Parlance Labs provides consulting and education for teams building AI products with large language models. It works with companies that have moved beyond prototypes but need repeatable ways to diagnose failures and decide what to improve next. Its AI Evals for Engineers & PMs course teaches evaluation methods, while its evals skills give coding agents reusable workflows for auditing evaluation pipelines, reviewing traces, generating test inputs and validating automated evaluators.

Founder Hamel Husain runs the business, drawing on machine learning experience at GitHub and Airbnb. Repeated evaluation problems in consulting engagements led him to develop the course with Shreya Shankar. His published Critique Shadowing approach starts with domain experts making pass/fail judgments and explaining their reasoning, then uses those examples to develop and test an LLM judge. The emphasis is on evaluation criteria grounded in a product’s actual tasks and human judgment.

In 2026, the company’s website reported 4,500 engineers and product managers had taken its course. Separately, Husain and Shankar described experience helping 50+ companies in their joint publication on evaluation skills.