A small team of experts that does its own work

My name is Rob Churchill. I have been building data, machine learning, and AI systems for a decade, and I have run Churchill Software as a consultancy for the last five years of it.

The company is small on purpose. We take a limited number of engagements at a time, which means the people you talk to are the people writing the code. Our team consists of a small number of senior engineers with strong backgrounds in data and AI. We all know how to solve business problems from a technical point of view, and we communicate in plain English, no acronyms, no jargon, no hand-waving.

The projects that come to us

Most of our projects come to us from a similar archetype of company. Someone has a great idea for a product, or has potentially built (or vibe-coded) something that works. There is a gap between an idea or a proof-of-concept, and a production-worthy system that their business, or their clients, can rely on. That gap can exist for many reasons: It can be data no one trusts, definitions that were never agreed on, or having no way to tell whether a change improved the outcome. It can be a lack of engineering resources or subject-matter experts.

That gap is interesting work, and it is chronically underserved. In large software-centric companies, that gap is bridged by armies of software engineers, product managers, and designers. Most companies, even large companies, do not have these armies to deploy. Sometimes, our clients have their own engineers in-house, who may just not have the necessary AI or data expertise. Sometimes, our clients have no engineers. Large consultancies staff their projects with whoever is available, and product teams try to absorb the extra work alongside their roadmap. We are fully focused on bridging this gap, and we do it surgically. Our small team of senior engineers all have the necessary expertise to bridge the gap on their own, through solid engineering and communication with our clients.

Scoping, and saying no

We would rather scope a small piece accurately than a large one optimistically. AI and Machine Learning projects involve a good deal of research and development, rather than straightforward implementation. As a result, most of our projects start with something narrow that either proves the approach or tells us it is wrong, and grow from there if it is working.

Our philosophy is that selectivity gains trustworthiness. We always tell you when we think you should not do something, including when that means a smaller project, or no project, for us. We frequently advise clients not to pursue a project when we feel that it is not in their interest to do so. That has cost us work and it has also brought back more of it.

Founded

2021. Registered as Churchill Software, LLC.

Team

Small and senior. Specialists brought in when a project genuinely needs one.

Background

A decade of data and ML engineering, and a Ph.D. in computer science from Georgetown.

We work in

OpenAI and Anthropic APIs, open-weight models, SQL and NoSQL, data and ML pipelines, AWS.