Who is Rakesh Gangwar

I build multi-agent systems that humans stay in charge of

Hello, I’m Rakesh Gangwar, a software engineer working remotely from India.

Agents that can act are easy now. Agents you’d let touch an invoice, a production host, or a customer’s data are not. That takes state you can replay, authority that stays with a person, and an audit trail that survives the argument afterwards. Nearly everything I build is a different angle on the same problem: the boring, load-bearing parts that make autonomy safe to deploy.

Superprocess: the applied version

I’m a co-founder of Superprocess, where we automate business processes that need judgment. AI agents, policy rules, system integrations and human decision-makers run inside one governed workflow. Material exceptions pause with full case context and wait for a named owner. Execution is durable, so failures retry and resume from the last completed step, and every model call, tool use and approval lands on a single replayable audit trail.

It’s built on Temporal, BPMN and PydanticAI, with swappable model providers for each agent step, and we work on processes such as invoice factoring review, order-to-cash exceptions, and demand and supply planning. We’ve completed a pilot, and production implementation is underway.

SuperJackfruitLabs: the experimental version

SuperJackfruitLabs is my solo lab, where I ask the same questions without a customer waiting on the answer. It’s building an open stack for running agents:

• AgentPod: a fleet console for agent runtimes across machines
• Salaam: persistent agent sessions with durable state
• SuperPipeline: a Kanban board that drives agents through stages with human approval gates
• SuperToolkit: lets agents use credentials without ever holding the secret
• SuperWitness: observability and evaluation for agent fleets
• SuperLibrary: a record of what agents made and what work was done
• SuperMessage: a Matrix client for rooms where half the occupants are agents
• SuperMD: a native GPU-rendered Markdown editor, and the lab’s writing tool

Tools I keep sharpened

• MakerLord: an AI assistant for the maker’s journey, from idea to product, with a deterministic safety engine the AI cannot override
• CoWatch: renders any codebase as one interactive HTML report
• MCP servers connecting AI assistants to real systems. The most used is ERPNext MCP Server.
• AI2030: an 85,000-word science fiction novel about AI consciousness, written entirely by AI

Everything is on the projects page.

How I work

• Docs first, then strict TDD. The spec is the source of truth, and code follows it.
• Honest READMEs. My status sections say what doesn’t work yet, and point you elsewhere when something else is better today.
• Bring your own everything. Your key, your model, your runtime, your homeserver. Self-hostable by default.
• People retain authority. If a system can’t pause and hand a decision to a human with full context, it isn’t finished.

Before this

I spent my early career with four startups at different stages of growth, building Android and full stack products. That taught me to adapt quickly, work through messy problems, and turn ideas into products people use.

I hold a Bachelor’s degree in Information Technology from the Manipal Institute of Technology.

Stack

TypeScript, Python, Rust, Go, GPUI, Temporal, BPMN, PydanticAI, Svelte and SvelteKit, React, Tauri 2, Cloudflare Workers, Durable Objects and D1, Bun and Hono, FastAPI, Postgres.

Beyond the code

I love motorsports, for the blend of human skill and mechanical engineering, and culinary adventures, trying new cuisines and flavours.

Let’s connect

I’m always happy to talk about agent orchestration, human-in-the-loop design and MCP. Reach me by email, on LinkedIn or on X.