AI can write the code. I work on making it safe to ship.
I lead Enablement Engineering at ANZ in Melbourne, the platforms, test infrastructure and CI that banking teams ship through. Lately that means figuring out what changes when AI agents join the delivery path: evaluation, code review, sandboxing, and the points where a human still has to say yes.
What I work on
- 🤖 AI agents in the SDLC: coding agents in CI, agent evaluation and tracing, MCP tooling
- 🛠️ Developer platforms & enablement: CI/CD, release engineering, golden paths, local-first dev tools
- ☁️ Cloud & event-driven systems: GCP, AWS, Kafka, Pub/Sub, gRPC
- 🧪 Testing at scale: end-to-end, API, mobile, contract and event testing frameworks
Featured projects
| Project | What it is |
|---|---|
| Restura | Local-first API client for HTTP, GraphQL, gRPC, WebSocket, Kafka, MQTT and MCP. Web, desktop and Docker. restura.dev |
| deep-agent-action | Mention @agent on an issue or PR and get a plan, edits, a PR or an inline review, all running on your own Actions runner |
| openapi-go-mcp | Generate Go MCP servers from OpenAPI 3.x / Swagger 2.0 specs |
| HybridTestFramework | End-to-end testing for web, API, cloud, events and security (Java) |
| JARVIS | AI-powered CLI for API test generation and traffic inspection |
| pubsub-emulator | Local GCP Pub/Sub emulator with a web dashboard |
Recent writing: more at dipjyoti.dev/blog
- Evaluating AI agents like software
- Containing AI agents at the OS level
- From OpenAPI to MCP with Go
- RAG is not vector search
Certifications: GCP Professional Cloud Architect · GCP Professional Cloud Developer · AWS Solutions Architect Associate · Google Generative AI Leader · GitHub Actions
Find me: dipjyoti.dev · LinkedIn · X · Medium · RSS





