Cover Letter for Volunteer: BYOM: Adapter for AI and LLMs
Himanshu Jha
I am Himanshu Jha, applying for the LFX Mentorship Program – Term 3 for the project “BYOM: Adapter for AI and LLMs” under the organisation CNCF: Meshery.
I found out about the LFX Mentorship Program while being part of the Meshery and Layer5 open-source community. Over the past few months, I have contributed to several repositories, joined contributor meetings, and worked with contributors on issues, pull requests, and documentation. Through this experience, I became increasingly interested in how Meshery can integrate with different AI models and, more importantly, how the complete flow works — from a user providing their own API key to an AI model generating a validated Design that can be used on the Meshery canvas.
This project is a great fit for what I enjoy working on: cloud-native tools, Kubernetes, AI/LLM integrations, RAG pipelines, and technical documentation. I am particularly interested in understanding how different AI providers and models can be integrated into Meshery in a way that is flexible, reliable, and easy for users to work with. The BYOM approach is especially interesting to me because it gives users more control over which models and providers they use while keeping the experience within Meshery.
What excites me most about this project is the combination of technical and practical problem-solving. Working on an adapter that can connect different AI/LLM providers with Meshery requires understanding APIs, authentication, model interactions, structured outputs, and validation. At the same time, the final experience needs to be simple enough that a user can provide their own model credentials, describe what they want to build, and get a usable Design on the Meshery canvas. I would really like to contribute to making this workflow reliable and easy to understand for users.
As a Layer5 community member, I have contributed to several Meshery repositories, including:- meshery/meshery
- layer5io/docs
- layer5io/layer5
- meshery/schemas
- https://github.com/meshery/meshery/issues/10878
- https://github.com/meshery/meshery/issues/20795
- https://github.com/meshery/schemas/pull/1059
- https://github.com/meshery/schemas/pull/1125
- https://github.com/meshery/meshery.io/pull/2868
- https://github.com/layer5io/layer5/pull/7902
Beyond writing code, I have also taken part in discussions on the Meshery discussion forum. One such discussion was around the Makefile build for the Meshery.io repository failing. After discussions over internal Slack, this eventually led to the following pull request.
Another instance where I helped detect a bug related to Design import in Meshery, which was addressed through the following pull request. I have also helped triage issues, reviewed documentation, and published my own Design on Meshery. These experiences have helped me understand how Meshery is structured and, more importantly, how collaboration, discussions, testing, and good documentation all contribute to building a better open-source project.
Kubernetes and cloud platforms (AWS)
DevOps and systems administration
YAML and JSON configuration
Git and GitHub workflow management
CI/CD and automation pipelines
Python scripting and automation
Documentation-driven development
- AI/LLM integrations
- RAG pipelines and AI application development
These experiences have grown my interest in systems that help developers better understand, create and control their deployments, which is exactly what this project aims to do.
Through this mentorship, I want to learn more about how AI and LLM integrations can be designed within a cloud-native platform like Meshery, how different model providers can be abstracted through a common interface, and how to create high-quality tutorials that make complex topics easy to understand. I am looking forward to working with the mentors and making contributions that will keep helping the Meshery community long after the program ends.
Thank you,
Himanshu