Where ambitious ideas become working systems.
Twenty-eight years across the first Nike+, data infrastructure for a Microsoft Research quantum program, Fortune 500 cloud architecture, and Retrieval-Augmented Generation systems running in production inside real enterprise permission models. Founded and run by Geoffrey Roth.
A production RAG system earns its place by turning your corpus into answers people can trust and trace.
The value comes from the system around the model: retrieval designed for an accumulated corpus, authorization that respects the compliance boundary, and evaluation grounded in the questions users actually ask.
RAG is one expression of twenty-eight years in full-stack, platform and cloud engineering. The same principal leads every engagement, with experience spanning New York and a physics lab in Delft. Banyan works worldwide.
Thirty years of inconsistent documents across four enterprise systems, governed by a security model that decides who sees what. Two years in production, presented anonymously while the engagement is active. The defining work is the retrieval, identity and authorization architecture around the model.
A large enterprise with tens of thousands of internal documents spread across a document management platform, plus a service management system carrying knowledge-base articles and years of incident history. Staff ask questions in plain language and receive answers grounded in the actual corpus, with citations back to the source.
Built on the Microsoft stack end to end. Scheduled jobs pull documents out of the source systems into object storage, where an indexing pipeline makes them searchable. A relational store backs the application itself, holding conversations and the excerpts behind every answer. Retrieval selects the evidence; the LLM turns it into a cited answer.
A shared open agent protocol serves the web application, a bot inside the chat platform staff already use, and other authorized clients. Single sign-on runs across two federated identity providers, and that identity governs retrieval from the first query.
Each answer draws from material authorized for that person's identity. Authorization is enforced inside retrieval, per user and per document, using the role model the business already runs on. The identity that signs a user in through either federated provider governs which passages retrieval supplies to the model.
Planning a system around enterprise knowledge, identity and retrieval?
Start a conversation →From consumer platforms to quantum data and enterprise RAG, the assignment has stayed consistent: make complex technology tractable enough for other people to build with it. That has been the job since 1998.
On the team that built the first Nike+, before connected fitness was a category. It took the Cyber Grand Prix at Cannes. Also the first e-commerce Nike store, and de-facto tech lead across nikerunning, nikewomen, Nike Be True and AF1.
For Microsoft Research, through a Seattle-based consultancy: the Python CLI that pulled run data off the fridges cooling the processors near absolute zero, and the Azure platform that ingested it, so labs on different continents could share, visualize and annotate each other's runs. On-site with the physicists in Delft.
Consulting architect, then Principal Engineer. Full-stack and cloud architecture for clients from seed-stage startups through the Fortune 500, and mentoring the engineers who built them.
Retrieval-Augmented Generation (RAG) and Azure architecture for organizations whose accumulated knowledge is the product. Concurrently finishing an MS in Computer Science at Georgia Tech, specializing in machine learning.
Frontier research rests on engineering that captures its data, moves it and makes it usable. On both of these programs the science belonged to physicists and molecular biologists, and I was on the teams that built that engineering. The links introduce the programs; my contribution to each is stated directly above them.
Microsoft ran experimental quantum labs at Copenhagen, Delft, Sydney and Purdue, pursuing a qubit whose error resistance would come from physics rather than software. The labs shared an open-source measurement framework called QCoDeS, an acronym for the three universities that built it.
Each processor ran inside a dilution refrigerator, which the physicists called a fridge, holding it a few thousandths of a degree above absolute zero. Every run produced measurement data that had to be captured with its full instrument state intact, then somehow made useful to researchers in another country.
Through a Seattle-based consultancy, I built the Python CLI that pulled run data off the fridges, reading it via QCoDeS. Then I architected and implemented the Azure platform that ingested it, so labs across the program could share, visualize and annotate each other's runs. Some of that work was done on-site alongside the research team in Delft.
Structurally, it is the same work I do now: specialized data, made findable and useful.
The Molecular Information Systems Lab, a joint Microsoft Research and University of Washington effort, encodes digital files into synthetic DNA. The appeal is archival: extraordinary density, and a read technology that stays relevant for as long as biology does.
The team set a 200 MB storage record in 2016, demonstrated random access across more than thirteen million DNA strands in Nature Biotechnology in 2018, and built the first fully automated write-store-read system in 2019.
Through a Seattle-based consultancy, I overhauled the existing encoding pipeline: the layer that turns binary data into nucleotide sequences a synthesizer can actually manufacture.
Banyan is a small, principal-led practice. You work directly with the people who scope, design and build the system.
I write code throughout every engagement and stay accountable from the first working session through handoff.
When an engagement needs specialists, I bring people I have worked alongside for years. The same names, engagement after engagement. You meet them before they start, and you keep working with the same faces throughout.
Banyan takes deliberately few engagements so each receives sustained principal attention. I confirm availability and timing on the first call.
Tell me what you're building, where the knowledge lives, and who needs access. I'll reply with a direct assessment of fit and the most useful next step.
