About
The studio
FriendMachine is an independent AI and data studio. I started it in 2020 to hold consulting work and side projects. It's since grown into a home for the products I build and the research that informs them.
Most of the problems I work on involve recovering structure from incomplete or noisy inputs: sparse sensor networks, unsynchronized video, partial location data, distributed agents without global state. The domains differ; the shape of the problem stays similar.
Joshua Winters
I studied computer science and ecology at the University of Michigan. The combination wasn't a plan. I just kept finding that problems in one field had better formulations in the other.
From 2015 to 2020 I was at IdealSpot, a seed-funded B2B SaaS that used geospatial and consumer data to help commercial real estate teams reduce site-selection risk. I joined as lead backend and data engineer and became CTO, running engineering, data, product, and frontend across a team of nine. The product sold as both an application and a data API; the assets were acquired in 2023.
Since 2020 I've been Director of Software Engineering at Sensei Ag, a controlled-environment agriculture company co-founded by Larry Ellison. I wrote the first prototypes, then built the operations software team and set the architecture. Working in an R&D partnership with Applied Invention, Danny Hillis's engineering studio, we built the systems that run daily operations on the company's farms.
FriendMachine is where I build the things I want to build.
How I work
This is a one-person studio. If we work together, you work with me directly.
The best fits are problems with messy data and half-formed requirements, where someone needs to think carefully before writing code. Consulting, prototypes, research collaborations, early-stage technical partnerships. If you have a clean spec and need hands to build it, I'm probably not the right choice.