Talk to us

Oatlab is one person: me, Robert. That's not something to work around; it's the point. When you get in touch, you reach the person who does the work.

What to expect

• One person, start to finish. You email me, I reply. You work directly with the person writing the code, with no account managers, no handoffs, and no waiting on a committee.
• Comfortable on a team. I do my best work alongside other people. I can plug into your team, or bring you into mine; either way you get a collaborator, not a black box.
• New to business, quick to move. I've spent my career shipping software; running the business is newer to me. You get plain pricing and a fast answer, with no sales process in between.
• AI is new for everyone. Nobody has decades of experience here. I've been building AI software full time since spring 2026, early, and shipping.

What I do

I provide direct AI services: I work on the models, the plumbing, and the product myself. That covers local and on-device AI, model integration, retrieval over your own documents, tool use, and agents. I work with clients in the Washington, DC area and remotely.

What I'm available for

• Contract work: project-based or ongoing.
• Employment: open to joining a team full time.
• Bronco: questions, or bringing it to your team or organization.

My AI Story

I got my introduction to AI while working at a company called Endgame, where I was privileged to work in Agile/Scrum teams staffed with Data Scientist members. Our software processed data with AI mostly to categorize it. I also worked on natural-language-processing (NLP) based chat-bots at Endgame. I had the chance to put serious effort into trying to make open ended text based assistants in 2015, an experience I am grateful for now that LLMs have made the exercise a joy.

I was also introduced to Bloom Filters at Endgame: an evocative data structure which represents a 'Set' and which is highly space efficient and where false positives are possible (but false negatives are not.) These can be used for Anomaly Detection, a key part of the Endgame offering. Later, while working at Elastic after they acquired Endgame, we would use a variant of the Bloom Filter: HyperLogLog Plus, to estimate the cardinality of data.

At Endgame I worked around the clock figuring out how to make 3D visualizations of process graphs for our product. I had to understand the process model, graph visualization and interactive graph visualization design, and 3D modeling and 3D interaction design plus I had to ship it. The first version used WebGL which ended up causing support issues in the field, so my next version involved designing 3D renderer made in CSS. Especially to achieve the 3D modeling and zooming-around camera animations, I learned the basics of linear algebra and the quaternion number system.

While at Elastic I learned all about Information Retrieval, but also how to design software that exposed a huge amount of functionality to users while also supporting long running projects at massive scale.

At Prequel I got the chance to do a lot of roles on a small team again, and I got to spend time with Three.js again, Monaco (the library sharing the core behind Visual Studio Code) and most of all coding with Cursor's agent mode as my primary tool.

Building Bronco

In Spring of 2026 I decided to focus full time on building AI tools. I am confident in the transformative power of AI, and particularly I am passionate about human-scale, creative, language-based tasks.

I explored creating cloud-hosted agents and generative art websites before deciding to focus on local AI. Local AI allows me to put a skilled human user in front of powerful desktop software with private and secure AI assistance, all on-site and cost controlled.

Bronco is the AI product I want to use, and the AI tool I want for my family. Bronco is a tool for mastery. It's a tool for learning safely. It's a tool you can rely on in the long run.

I recommend you try it with your WiFi off or on Airplane Mode at first. It's relaxing to have Bronco's capability with total privacy. I hope you'll agree.