AI developer recruitmentClients in the UK, Europe and USA
If you are building products on top of models, you need an engineer who can make retrieval, evaluations, agents and API integrations reliable. An engineer reads every CV first, so you only interview the ones who can.
01
AI experience is the easiest thing to add to a CV and the hardest to check. A side project that calls a model and a production retrieval system look the same on paper. So every CV gets a second read from your most technical person, and the difference often only shows in the interview.
02
You need CVs that pass engineering’s interview, so you can send a shortlist it trusts. An engineer reads every CV before it reaches you. Fewer CVs, but better ones.
58%
of HR and recruitment leads who use agencies say the main reason engineering rejects agency candidates is that the CV looked good, but the person couldn’t do the job.
A developer or CTO who works in the stack interviews every candidate. Each of these is a question you would otherwise only answer in your own interview.
Whether they can design retrieval, reason about embedding and reranking quality, and separate retrieval failures from generation failures.
Whether they build meaningful offline and online evaluations, so a prompt or model change is a measured decision and regressions are caught.
How they handle model failures, latency, cost and confidential data, how they limit what an agent is allowed to do, and how they stop low-quality output from silently reaching users.
We find model-training specialists too. For those roles we test architecture, training dynamics, fine-tuning methods such as LoRA and QLoRA, evaluation for regressions and catastrophic forgetting, and serving on GPUs.
They understand what happens between the prompt and the answer, and can investigate a bad output instead of rewording the prompt until it goes away.
They can explain how they verify generated code and model output, protect confidential data and stay accountable for what the feature does in production.
They resist overstating what a model can do, and say plainly what they have built, what they have not and what the evaluation does and does not prove.
We recruit AI engineers who build products on top of models, and machine learning, LLM and AI research engineers who train and adapt them. Every candidate is screened on retrieval, evaluation, model integration and production reliability, with training depth assessed where the role needs it.
Either way you get about 3 to 5 screened CVs a week, per role.
We do
You do
We do
You do
“Metamindz helped us recruit our now CTO, reviewing CVs as they came in and then putting the CTO through a thorough 90-minute live 360 technical test. Feedback from our now CTO of the test was glowing and comms from Metamindz throughout the entire process were excellent. Highly recommend!”
Rupert Boddington
Co-Founder, Pockla
“Worked with Lev and Max. We worked with MetaMindz to help us interview and assess software engineering candidates. They were professional and great from the outset and throughout, great speed too. Would 100% recommend and we will be working with them again.”
Alexander Bushe
CEO, Fuller Focus
“We found it really easy working with Lev and the team to hire for our lead developer role and despite it still being early days, our new hire is making an impact already!” source
Billy Whiffen
Google review
Built the AI front end, set up a design system that removed the need for Figma, and architected analytics properly.
Impact as described by Billy Whiffen, eLocker.
£5,000
Fixed fee for your first role
15% to 17%
Commission on each further hire
Deep technical interview (Route 2)£300 an hour
A 1.5 to 2 hour live-coding and technical grilling session per final candidate.
All fees exclude VAT.
Tell us about the roleEngineers who build products on top of models, covering retrieval, evaluations, agents and model API integration. We also recruit machine learning, LLM and AI research engineers who train, fine-tune and serve models.
Yes. We test data preparation, training objectives, optimisation, distributed training, full and parameter-efficient fine-tuning, evaluation, failure analysis and the trade-offs involved in adapting foundation models.
Our assessment can cover RAG, embeddings, reranking, model evaluation and MLOps, and for training roles PyTorch, TensorFlow, JAX, Hugging Face Transformers, supervised fine-tuning, PEFT, LoRA, QLoRA, RLHF, DPO and distillation.
Give us a few details about the role and your hiring plans. We'll assess your needs and get back to you with a practical next step.
Listed for reference. We use these to scope the brief with you. A keyword on a CV is where screening starts, and the interview decides whether the candidate can do the work.
Teams usually bring us in for: A product team adding its first AI features. An AI product that needs retrieval and evaluation to hold up in production. A team developing or fine-tuning its own model.