Hey, I'm Obed. I have eight years of experience, most of it on the backend, working on background jobs, data integrity and the kind of failures that only show up once real traffic hits a system. Most recently I was at Search Atlas, where I worked on the Django and Celery backend behind an AI content platform and built a multi-agent LLM runtime for routing calls across different model providers. When the team's remit expanded beyond backend engineering, I picked up React and Next.js work too, along with release verification and the RCA write ups nobody enjoys writing but someone has to.

Before that I led engineering at Crop2Cash for just over three years. We redesigned the database and reworked the codebase behind a USSD application, which helped drive a 1,336% increase in user acquisition over a year, and built a multilingual AI voice advisory line that let farmers call a toll-free number for agricultural advice without needing a smartphone. That project took the extension agent to farmer ratio from roughly 1 in 10,000 to 1 in 1. I've also been Head of Engineering at BottleKing since late 2024, rebuilding their Node backend into microservices on AWS, work that's helped grow orders roughly 4x and revenue roughly 6x so far.

One thing I'm oddly proud of is a set of end-to-end verification rigs I built at Search Atlas, using Dockerised services, real Celery workers and actual WordPress and Wix sandbox accounts, because our unit tests kept passing on code that quietly broke in production. Mentoring junior engineers has taught me the same lesson from the other side, that it matters more to explain the reasoning behind a decision than to just point at convention.

Tools I reach for most: Python, Django, DRF, Celery, LiteLLM, the OpenAI Agents SDK, Postgres, ClickHouse, Redis, React, Next.js, Docker, AWS, GCP and Azure.