// What I offer
This is roughly how an engagement unfolds: a conversation, then an audit, then a first build, then a standing slot in your week. Most clients move down it over a year. You do not have to start at the top, and plenty of people skip straight to the bottom.
You want to know whether I am the right person for this.
I ask a lot of questions and tell you what I would look at first. If I am not the right person, I will say so, and often point you to someone better.
You have several ideas and no clear priority, or a direction agreed and no idea how to build it.
I run a discovery audit to settle which problem to back, or a planning audit to map how you would build it, ending with a concrete roadmap.
You have the plan agreed and nobody can tell you whether the hard part actually works.
I build a first version myself, usually against offline or stand-in data rather than your live system. It is not production code and is not meant to be: you get to try the idea before committing a team to building it properly.
You have a problem that will take months to get right, and nobody senior enough in-house to own it.
I take a standing slot in your calendar every week and stay reachable in between, anywhere from a couple of hours to a full day. Your engineers do the building and learn the field while they do it, and I have taken teams from no machine learning at all to shipping it in production. I work the hard parts alongside them and explain the reasoning as I go, so it stays with your team after I leave.
Also, less often
A paper, or a vendor's claim, is about to become a roadmap commitment. An independent read on whether it is real and whether your team can build it.
The same question asked about a whole company, for the investor about to fund it: the technology, the team behind it, and what would have to be true.
Robotics is where I started: electronics, mechanics, control, code. I do not build hardware day to day now, but it means I can work with your hardware and software teams at once, without things getting lost between them.
Planning or scaling automation in a chemistry lab runs through chemobotai.com, a partnership with Daniel Salley, ex-head of engineering at Chemify.
“Our dev team had very little machine learning experience. He coached a team of 3 developers on machine learning [...] Our first solution has recently gone into production and we are very happy with the results so far. He has proved invaluable as I suspect we would have ended up spending lots of time trying to tackle problems that were not suitable, too complex for our level of experience or approaching them in the wrong way.”
Senior expertise on a part-time basis, over months rather than in a single burst. You get someone senior in the room without committing to a full-time hire. I typically work with clients between one hour and one full day a week, over several months, to upskill their team and develop their projects.
Check that the expertise actually matches your problem, that there is relevant experience behind it, and that there is a track record on similar problems. Google my name, and read the credentials and testimonials pages on this site.
With a call, or a visit to your office, where I ask a lot of questions. For teams with no prior AI experience I usually suggest an audit first, then a first project with a small team at one or two hours a week, and we decide the next step from the result. If you already know what you need, I also work asynchronously on research problems, act as a part-time CTO, or pull in other experts from my network.
Invoiced monthly for the time actually delivered and agreed in advance. There is nothing to pay for a week I am not working. In some cases I can work for equity. I sign NDAs as a matter of course and carry professional indemnity insurance.
Often, yes. For UK companies, qualifying external consultancy spend on genuinely novel technical work can attract relief, which describes most of what I am hired for. Worth putting to your accountant before you budget.
