Harsh Kumawat
I’m Harsh Kumawat, a software engineer focused on AI training, LLM evaluation, and coding-agent assessment, with strong hands-on experience in repository-level debugging and automated failure analysis.
About
I’m Harsh Kumawat, a software engineer focused on AI training, LLM evaluation, and coding-agent assessment, with strong hands-on experience in repository-level debugging and automated failure analysis. I enjoy designing and improving benchmark tasks and tests so evaluation results are reliable, especially in terminal and software-engineering workflows.
Alongside my AI work, I’ve built and optimized production backend systems for media and OTT platforms using Java microservices, Dockerized environments, and dependable engineering practices. I also contribute to open-source C++ projects, where I’ve fixed memory-safety issues and added regression tests to improve long-term stability.
- 2 years' experience
- Joined November 2026
Industry experience
- Software & Internet
- Media & Entertainment
- Computers & Electronics
Work experience
Improved ingestion throughput for an OTT media platform by removing redundant API calls and adding asynchronous execution, increasing media ingestion from 1,000 to 3,500/day and non-media from 4,000 to 7,500/day. Built and debugged backend workflows for metadata aggregation, VOD, channels, EPG, SKU management, content packaging, and access policies, working with Java microservices and production debugging/optimization.
Developed PayTV workflows for Set-Top Box systems, building gRPC microservices and BPMN workflows. Validated APIs with Postman and wrote Python automation scripts to support testing and workflow development. Worked with a production-style stack including Docker, AWS, and messaging/integration tooling.
Education
Hire Harsh Kumawat today
Post a job and invite Harsh, or contact them directly. No hidden commission.