Deployed at Brain Co.

We’re an AI product engineer and AI/ML engineer in the rapidly growing deployed team at Brain Co.

Over the last year, we’ve shipped AI products into production across a range of complex environments, including government services, petrochemical companies, and healthcare organizations.

We’ve learned a lot from being embedded with customers, and want to share some of our learnings to help demystify what being a deployed engineer at Brain Co. looks like. 

‍

It’s 40°C outside in this coastal Gulf country. The air is shimmering over the hot asphalt as our teammate traverses a large chemical plant campus on foot. In the distance, he sees a rapidly approaching Land Cruiser. A client he's been working with had spotted him in the distance, and now insists on driving him.

In the air-conditioned truck, the conversation got personal. The client wanted to know what AI meant for his son's future – what should his son study, whether their jobs were at risk. Our teammate told him he didn't have answers, “just opinions”, and to take them as that. From that point on, the client trusted him enough to reach out directly for almost everything from bugs, to feature requests and platform walkthroughs. Not because he was the most technical person on the project, but because they'd had a real conversation.

That's one of countless small interactions that shape how our projects move forward as deployed engineers. Brain Co.'s mission is oriented around transforming governments and the world’s most important institutions. Though it sounds abstract and daunting, on the ground, it looks like showing up and working alongside our partners every day. A lot of the job is just being there in person, sitting beside users while they walk through their painful workflows, walking over to an engineer's desk to unblock a data request, or chatting by the water cooler about houseplant care (a tip we learned from a customer: try talking to your sad house plants. No, really!). All of those interactions help us foster much deeper empathy with our clients and allow us to build better products. 

We’ve learned that real transformation, even as AI progress accelerates, takes patience, trust, and genuine relationship-building to achieve.

‍

That’s nice, but what do you guys actually do?

We both came in expecting the job to be about 70% engineering, 20% client management, and 10% everything else. On average, this is true, though in reality, the ratio shifts project to project, person to person, and even week to week. Some weeks are all building, while others are knee-deep in user research brainstorming valuable use-cases. 

A randomly sampled week from our calendars: sprinting to ship a clinician workflow PoC Monday and Tuesday, workshopping a multi-tenancy deployment design on Wednesday, flying onsite Thursday to unblock a data request, and optimizing our medical coding model the rest of the week.  Another week has 2 days of scoping sessions, one day of user shadowing and using those client inputs to start building a solution for the rest of the week.

The engineering work spans product surfaces, enterprise integrations, and deployment architecture. Much of our effort goes into building representative evaluations that reflect the messiness of production data. Reaching the accuracy levels required for real-world performance often means experimenting with the latest research and doing novel work where standard playbooks fall short. We spend a lot of time identifying the subspace of meaningful errors, hunting down long-tail failures, and iterating until our systems consistently clear this very high bar.

We've come to love moving across deep engineering work, product judgement, and the trust-building required to ship in real institutions. Each week feels different, but our mandate remains consistent: to make our projects successful while owning the full processes and outcomes.

‍

The team behind the work

When your week splits across many contexts, having a stable home base helps. Many of us live close to our regional office, and the culture there is tight-knit in a way that makes the constant context-switching sustainable. We grab meals together, clash over chess and pool battles, and can’t miss the Arabica coffee breaks that are often highlights of the day.

Our team reflects the institutions we work with. In any given week, we work with partners across different languages and functions, in Arabic, English, Urdu, Spanish, and others – wherever they feel most comfortable. Our engineers come from almost every corner of the world. The people who thrive here are curious, adaptable, and genuinely interested in the places and organizations they work with.

‍

En Fin

We’ve learned that trust heavily compounds. A first meeting with a partner might just be coffee, dates and conversation, with the project discussion coming only after they’ve gotten to know you. But once trust is established, partners start to reach out directly when something breaks, advocate for the product internally, and tell us honestly when something is not working. Those relationships often become more than typical vendor-client dynamics, and they make our products better in ways that are hard to quantify but easy to feel.

This trust is the underlying infrastructure of deployed engineering at Brain Co. We're building it into and alongside our products. As deployed engineers, we see ourselves as a bridge between our company and our partner organizations, between engineering constraints and ambitious transformation visions, and between cutting-edge technology and legacy enterprise processes. Most days, that looks less like grand strategy and more like showing up and doing great work, alongside incredible teammates and the people we are building for.

We're hiring at Brain Co. If you want to build AI products in production with some of the world’s most important institutions, get in touch.

Thank you to Matias for contributing his deployed perspectives to this post!