Chat to a human for 30 minutes
You speak to a specialist talent partner about the role, your company, and the person you’re looking for. We’ll dig into the nuances of exactly what you want and don’t want. Mindy listens and writes a detailed brief.
Where the market is tight.
Two of the four hardest things to hire for right now are both in your brief. Plan for longer than 51 days, or look at step five, where I show you which line to loosen and what it buys you.
Who is actually out there.
- 1,800 engineers have been interviewed by a Wave consultant. 340 of those are ML-first.
- 62 clear your hard criteria — production ML, PyTorch, the years, London hybrid.
- Of those 62, 41 told us they want more architecture ownership in their next move. That is the part a job ad cannot find, and it is why the shortlist is short rather than long.
- Median salary expectation is £132k, inside your band. Six want more than £150k.
- 9 are actively looking. The rest are open but not searching, so they will never see a posted job. That is who I contact.
You fill in nothing. You come away knowing what your brief will actually get you.
Mindy finds exceptional people who fit
She searches the network of headhunted, interviewed talent for the best matched candidates.
Nobody has been contacted yet.
Mindy contacts everyone who fits
Mindy customises their outreach using what that person said they wanted. They click once if they are interested.
Hi Elena,
In July you told Alex you wanted to own the model architecture yourself, not tune someone else’s. This job is exactly that.
Solace AI build AI copilots for support teams. Four engineers, very little process. You would set the architecture.
Interested?
at Solace AI
This is one of the roles we have matched for you. We think it could be a strong fit, but have a read and decide for yourself.
About this role.
Solace AI builds AI copilots for support teams. You would be the fourth engineer and the first senior ML hire. There is no platform team, no PM layer and not much process. The model architecture is three people’s best guesses held together with evaluation scripts, and the job is to own it properly: decide the shape, build the serving layer, set the bar for what ships.
- Own the model architecture end to end — retrieval design, evaluation, serving
- Set the quality bar for the copilot and decide what “good enough to ship” means before the team scales
- Build the first serving infrastructure. There is no platform team to inherit
- Work directly with support leads at customer accounts on what the copilot gets wrong
- Make the build-versus-buy call on the model layer. Nobody sits above you on that decision
- Hire the next two engineers with the founders once the architecture has settled
A few reasons this one stood out.
In July you told Alex you had spent two years improving models you did not design, and did not want to spend a third. This is the opposite problem: four engineers, an empty platform layer, and nobody above you on architecture. The decision is yours to make and to defend.
Dana, the CTO, manages hands-off to a fault. Her words. She hands over a problem and steps back for a week. The last hire here failed because he waited to be told what to build, which is the opposite of how you described working, and worth knowing going in. The band sits around your £135k with equity on top at Series A.
Solace AI in a paragraph.
Most support tooling summarises tickets. Solace is building a copilot that closes them: it reads the account history, the product state and the conversation, then drafts and takes the action instead of suggesting one. They deliberately did not build a general assistant. Each model covers one support domain, it is scored against tickets that actually got resolved rather than against human ratings, and it only ships when it beats the team it is helping. Eleven customers so far, all B2B SaaS. The next year is about making that architecture hold at ten times the volume, which is why the role exists.
- Two European seed funds and a US Series A lead
- Angels from three support and CX businesses, one of them a former customer
What you’d be walking into.
5 steps, roughly.
Come ready.
- Come with questions about how the four of them make architecture decisions today. Dana expects to be interrogated on it and rates people who do.
- Be ready to walk through one system you designed from an empty repo: what you chose, what you got wrong, what you would do differently.
- Dana manages hands-off and says so openly. Have an answer ready for how you work when nobody is checking in.
One click. Nothing to fill in. The one-pager is written from the briefing call, not the job ad.
Interested candidates appear here, ready to speak to you
Say yes or no. Say why you passed and Mindy updates the brief.
Every reason you give is added to the matchmaking brief. The next batch is better.
Change one line, get more people
Mindy shows you which parts of your brief are blocking people, and which unlock Mindy surfacing more exceptional candidates that fit.
Give us one job and see who we find. It takes a 30-minute call to start.
Book a callMindy continually benchmarks the network of exceptional talent and works with you where to unlock the rest of the market.