Olame Galien Muhindo

Co-founder at Shwary, settling tuition for national and international students.

AI and machine learning, across the whole stack — the model, the retrieval, the services, the infrastructure, and the screen someone actually taps. San-Francisco,CA & Phoenix, AZ 🇺🇸

about

I'm a Co-founder of Shwary, a platform for settling tuition for national and international students within East African countries.

My focus is AI and machine learning. The part I enjoy is how many hats it still takes to ship one thing: the model, what you retrieve, the services behind it, the infrastructure under those, and the screen someone actually taps. I've never been good at staying in one layer of the stack, and I've stopped trying.

I want to point that at problems where a working product changes something material for the person on the other end of it. Right now that means getting a family's school fees where they're going.

In 2025, I've been thinking about the following questions:

  • School fees are one of the largest recurring payments a family makes, and one of the worst served. Why is it still normal to queue at a bursar's office with cash?
  • An international student's first experience of a country is often a wire transfer that loses a real percentage to fees and exchange rates, and then goes missing for a week. Who does that money belong to while it's in transit?
  • A university reconciles thousands of partial payments across banks, mobile money, cards, and cash, by hand. What becomes possible for a school that can simply see who has paid?

In 2025, I built AI products. At Necta I architected an agentic RAG system that figured out who you should meet at an event and why, and owned the recommendation stack end to end: retrieval, prompting, ranking, the services behind it. At WIZNiche I built a multimodal SMS agent that takes the administrative work off small businesses, and rebuilt the core of their mobile app so the next three years of features had somewhere to live.

The pattern I keep returning to is that the model is rarely the hard part. The hard part is everything around it: what you retrieve, what you show a person, and what happens when it's wrong.

In 2023, I led engineering teams in crypto. I was technical lead at Velix Protocol on a multichain liquid staking system, then CTO at QuariLabs, where we shipped DeFi protocols on Ethereum, Starknet, and Metis moving around half a million dollars a month. I'd started on smart contracts two years earlier at EarlyNode in Aachen, writing auction and payment systems and cross chain validators.

Most of what I learned there wasn't about blockchains. It was that a small team with no playbook will move faster than a large team with one, and that the job of leading is mostly deciding what not to build.

In 2022, I worked on M-Pesa at Vodacom Congo. Eight million people in the DRC used it to move money, and I shipped features into that.

I grew up watching cash move in ways that were slow, unsafe, and expensive for the people who could least afford it. Mobile money changed that more than any technology I've ever worked on. It also meant a failed transaction wasn't a bug report, it was somebody's school fees. I've never worked on anything since where reliability felt optional.

In 2019, I started studying computer science in Kampala, having grown up in the DR Congo. I took every job I could find, mostly building web and mobile apps, and started writing at olame.codes and posting engineering videos on YouTube, explaining the things I'd wanted somebody to explain to me. I now live in San-Francisco, CA.

My favorite thing is still the moment a thing works for the first time.

If you want to chat, book a call.

Olame Galien Muhindo