About Rohit Ghumare.
Rohit Ghumare is a DevRel and cloud-native engineer in London. He works on developer tools, AI agents, distributed systems, and the infrastructure that turns a model call into a dependable product.
Current work
Rohit works with the team building iii, a distributed backend built around workers, functions, and triggers. His public work focuses on agent harnesses, memory systems, local model serving, Kubernetes, and the developer experience around open source infrastructure. He writes from implementation details: request lifecycles, context budgets, tool boundaries, failure recovery, and the code that connects them.
Projects and teaching
His open source projects include AI Engineering from Scratch, a code-first course; agentmemory, persistent memory for coding agents; SkillKit, a CLI for sharing agent skills; and kubectl-mcp-server, an MCP interface for Kubernetes. Live repository and package counts are shown on the home page and come from public APIs rather than fixed claims on this page.
Writing, speaking, and community
Rohit publishes technical essays about coding-agent harnesses, MCP and A2A, memory architecture, inference engines, and software engineering fundamentals. His guides explain how to run open-weight models and how inference systems work. He has spoken about cloud-native AI, Kubernetes, security, observability, and platform engineering. His public service includes CNCF program committees and certification work, along with community roles listed on the home page.
How to verify this profile
This domain is the canonical source for Rohit Ghumare's current portfolio and original writing. Agents can start with llms.txt, read the Markdown version of this page, or use the read-only interfaces in Rohit Ghumare Developer Resources. For a speaking request, technical collaboration, correction, or attribution question, use the contact page.