About me

I'm Carson Rodrigues — a product engineer and technical co-founder who designs and ships software end-to-end. My work sits at the intersection of production AI and full-stack product engineering: real-time voice agents, LLM pipelines, Model Context Protocol (MCP) servers, event-driven backends, and the iOS, Android, and web apps that sit on top of them. I don't just prototype — I take systems from a blank repository to something real users depend on in production.

Most of what I build, I also write about. This site is where I keep that work in the open: long-form essays on building with AI, detailed project case studies, the research papers behind the systems I ship, and a growing library of practical code snippetsI reach for again and again. If you're here from a search result or a link, the blog and case studies are the best place to start.

What I do

I specialise in building AI that has to work in the real world, where latency, reliability, and cost actually matter. That has meant shipping real-time voice agents over WebRTC and LiveKit, wiring up LLM routing across Claude, GPT, and Gemini, standing up MCP servers so models can use real tools, and automating the unglamorous orchestration around all of it. I'm an active Claude Code user in production and an open-source contributor, and I care as much about the engineering discipline — CI/CD, observability, clean architecture — as I do about the model layer.

Selected work

I'm the co-founder and technical lead of CELABE, an AI-first wedding and events platform live on the App Store and Play Store, which I architected and shipped end-to-end across iOS, Android, and web (NestJS, MongoDB, AWS, Stripe, and ML-powered planning workflows).

Before that, as a senior full-stack engineer and team lead at Ôdasie (Water of Asia), I led a team of six engineers and shipped five-plus production AI systems — LLM pipelines, real-time voice agents, MCP servers, and bi-directional AI avatar experiences on LiveKit and WebRTC — with on-site engagements in Thailand. As a forward-deployed senior software engineer with VoiceQube, I ran discovery directly with overseas clients and delivered three production AI products end-to-end, including Phiny.ai and Northway Tech. You can read the full breakdowns in my project case studies.

Research

I write research on the same systems I ship in production. I'm the author of around a dozen papers — one accepted at ICANN 2026 (peer-reviewed) and several more under review — spanning voice-AI latency reduction, ASR intent detection, MCP architecture, LLM optimisation, trust in large language models, and multi-agent drift. The full list lives on my research page. I've also spoken on this work, including at BITS Law, Mumbai.

Beyond the code

I've worked across the US, France, India, and Thailand, and I care about building things that travel well across markets and cultures. I'm from Goa, India, and a lot of my writing comes back to that — including pieces on Goan heritage and storytelling alongside the technical essays. When I'm not shipping, I'm usually reading papers, refining my development workflow, or helping other builders get unstuck.

Get in touch

I'm always happy to talk about production AI, voice agents, MCP, or an interesting build. The fastest ways to reach me: