Kusho.ai.

My role

I joined Kusho.ai straight out of college as the startup’s first hire. As a founding engineer, I helped build an AI-powered API testing platform and took ownership wherever the early team needed support, across engineering and the wider work of getting a startup off the ground.

Making API testing easier

I designed and implemented a testing framework for REST and GraphQL, with OpenAPI and Postman integration. The work achieved 95% test coverage while reducing test-authoring time by 75%.

The goal was to bring test generation into the tools and specifications developers already used, so they could spend less time writing tests by hand.

Learning from real API traffic

I built a JavaScript SDK with middleware that worked across frameworks to capture and analyze API traffic. It detected anomalies and generated test variations from production traffic, helping reduce manual testing effort by 60%.

I also added LLM-based test generation to Swagger UI, extending testing into another familiar part of a developer’s workflow.

The systems behind the AI

I engineered a distributed task-processing pipeline using Redis and RQ to support scalable LLM operations. This work focused on throughput and reliability alongside the test-generation features themselves.

Being the first hire taught me to connect technical decisions with the needs of the whole product, take responsibility beyond a narrow task, and follow through until the work was usable.