20 June 2024

Welcome Fimio Platform Alpha

We are glad to be back. For the last six months, we’ve been heads down building and are excited to share with you all what we believe is a truly transformative product and new way of shipping AI products.

You last heard from us when we said we were pivoting. We are glad to be back. For the last six months, we’ve been heads down building and are excited to share with you all what we believe is a truly transformative product and new way of shipping AI products.

The process of transforming code into a product built for others to use and derive value from is a team sport. At Fimio, we want to give those teams an easy, repeatable, straightforward way to collaboratively ship AI products. We have created a new paved path for taking AI models through to deployable AI products, on a unified platform for everyone involved in the project to collaborate and ship together.

Our opinionated workflow maintains and preserves the link from code to deploy for the entire lifetime of an application. With Fimio, AI engineers will not have to struggle to get their code to production, the infrastructure team will easily know if the build will work on their server, the product manager will be able to test and give their approval on the build, and devs ops can avoid the dreaded push to production with fingers crossed – and everyone will know explicitly that the product owner has given their signoff.

So why this, why me, why now?

Tl;dr: because I’ve felt this pain acutely. As a trained Computer Scientist turned Applied Artificial Intelligence (AI) Engineer I spent most of my time focused on writing algorithms, focusing intensely on the code and not really paying attention to what happens to that code when it leaves my Jupyter notebook. One consequence of this was that most of my code was written in a way that was not reliable, and not reproducible. If a software engineer or a devs ops colleague wanted to turn my code into a product, it would have involved a lot of back and forth because they would have to go on the journey of discovering all the implicit dependencies (code, machine, third-party APIs, data, and so forth) that my code needed to work as I envisioned it. Furthermore, if product people wanted to do functional testing on that code, it would have been next to impossible for them to do so.

I founded Fimio to apply machine learning to difficult problems, and realizing that what foundational tooling does exist for shipping AI is mostly focused on training AI and exclusively aimed toward AI engineers, the product we have come to affectionately call VCoW (Version Control over Workflows) began immediately taking shape. From my experience, I knew that shipping AI products involved multiple organizational functions from product management, to developer operations, to infrastructure engineering, and more. And all those teams had to collaborate to ship a product, and all those teams needed to interact with the process at different times. We knew if our goal was to make it straightforward and easy to ship AI products, we had to close the gap that existed between code and deploy phases.

Check out Fimio here

We can't wait to share the Fimio platform with the world! We’re currently in alpha, so join our waitlist, or if you have a hair on fire problem that we could put out, email me at omoju@fimio.xyz.

The Science of Deployment: Ensuring Reproducibility in Modern Software

As someone who earned my PhD in Interdisciplinary Computer Science and spent the last 10+ years working in machine learning, AI, and software development, I can confidently say: there’s a deployment problem across tech, and more broadly, there is a crisis of reproducibility in science. So what does that mean and why is it important? I’ve been thinking about that for years, and I’ve got answers.