Skip to content

Services

Software, ML engineering, and MLOps services.

Software design, scalable infrastructure, and ML engineering. Short, focused engagements — one to four weeks. Vancouver-based, working remotely across Canada and beyond.

Who this is for

No prerequisites. I've worked with people who have nothing but an idea, small businesses with no technical team, and startups that already ship code. What matters is a defined problem.

What I do

POC and MVP development

Turn an idea, proposal, or research project into a working version you can show, test, or ship. Most are one to four weeks. Typical stack: Python, FastAPI or Flask, React or Flutter for UI, GCP or AWS for hosting. I build with deployment in mind from day one.

ML and automation adoption

For businesses and teams exploring where machine learning, automation, or LLMs could help. I start with a short audit — what you do, where the manual or error-prone work lives, what's worth automating and what isn't. You leave with a prioritized plan and, if you want, a working prototype of the highest-value piece.

Includes agentic AI — systems that use models and tools together to automate multi-step workflows.

MLOps and research-to-production

For teams with working research code that needs to become a reliable, scalable service: containerization, CI/CD, model serving, monitoring, and the infrastructure that makes an ML system maintainable. Typical engagement: two to four weeks of focused setup plus a handoff.

Architecture and best-practices review

For teams that suspect their stack, tooling, or process is slowing them down. I review code, tooling, and development process; identify the bottlenecks; and deliver a written report with a prioritized next-step list. Delivered as a document, not a consulting deck.

Also covers software and ML system design, and advising non-technical founders on structuring and hiring into a technical team.

Perception data & aerial capture

Background in large-scale perception data platforms (imagery, lidar, video, sensor timeseries). I have capture hardware (sub-250g drone) for mapping, photogrammetry, and surveying — available when relevant, not bundled into the quote otherwise.

What I don't do

I'm an engineer, not a researcher. I build on proven tools — open-source frameworks, cloud platforms, pre-trained models — and focus on getting working software into your hands. The goal is always something you can run, test, and ship.

Train foundation models from scratch

I use pre-trained models and fine-tune when needed.

Build custom ML frameworks

Existing tools are mature enough for nearly every business problem.

Academic or theoretical research

I work on things that go to production.

Short, defined engagements

Every engagement starts with a defined problem and a measurable deliverable. Most projects run one to four weeks. Larger efforts get broken into sequenced sprints, each with a clear decision point before the next begins. Ongoing relationships work the same way — structured around outcomes, not open-ended hours.

Current industry experience

I bring active experience from the techbio space — hands-on with ML and MLOps work in production right now.

Start where you are

You probably have enough to start. You don't need a finished plan — figuring out what to build is part of what I help with.

How an engagement runs

  1. Conversation. Thirty to forty-five minutes. What you're trying to do, what you've got, what success looks like.
  2. Audit and plan. I review what exists and write a short proposal — scope, deliverables, timeline, cost.
  3. Build or advise. The focused work. Weekly check-ins or async updates, whichever you prefer.
  4. Handoff. Final report, code, and setup documentation — what you have, how it works, how to run it.

Start with a conversation.

Arrive with a problem, not a brief. The worst outcome is a short, honest conversation.

Start a conversation