Adding distributed tracing to AI Gateway: My LFX mentorship journey

My background and preparation Application strategy: Contribute first, apply later Deep product experience Proactive problem solving Why this approach is important Project core: Adding distributed tracing to AI gateway The problem to solve Project goals Design approach Important lessons from the development process Critical testing strategy lessons Key takeaways Community collaboration Project results and value Implemented features How to experience the distributed tracing feature Personal gains Advice for students who want to participate in open source LFX mentorship application strategy Project execution advice Conclusion Posted on October 20, 2025 by Zhengke Zhou CNCF projects highlighted in this post In today’s rapidly evolving AI landscape, effectively monitoring and debugging AI Gateways has become a critical challenge. This article shares my complete experience through the LFX Mentorship program, where I added OpenTelemetry distributed tracing support to kgateway’s AI Gateway functionality. From application strategies for LFX Mentorship to challenges and insights during project implementation, I hope this provides a valuable reference for students who want to participate in open source projects. Before applying for LFX Mentorship, I had already been exposed to [OpenTelemetry](https://opentelemetry. io/) during my internship, gaining foundational knowledge in the observability domain. More importantly, I participated in the Jaeger community’s development work to add Clickhouse support for traces ([PR #6935](https://github. com/jaegertracing/jaeger/pull/6935)), which gave me practical experience with distributed tracing. These experiences made me feel that the LFX Mentorship project about AI Gateway distributed tracing was an excellent opportunity to deepen my learning and contribute to the open source community. I know everyone gets excited when they see a project they’re interested in and can’t wait to apply. I adopted a different strategy: first deeply understand the project, actively participate in the community, and then submit the application. Instead of rushing to submit my application, I first went to actually experience the product involved in the project: [AI Gateway](https://kgateway. dev/docs/ai/about/).