Founding ML Engineer
Set the modeling bar for everything we ship into regulated work.
The role
You will lead the modeling work on our first wave of regulated deployments — from synthetic data and fine-tuning to deployment, evaluation, and post-launch retraining. You own the path from a customer eval to a production model that defends itself.
About Neuralcraft
Neuralcraft is a frontier AI studio building decision systems for regulated work — insurance, healthcare, audit, public safety. We take engagements end-to-end: research, models, product surface, and operations.
This is a founding hire on the modeling side. You will work directly with the founders on the first three customer deployments, define the engineering practices we'll scale into, and be a load-bearing voice in shaping the product.
What you'll do
- Own the full modeling stack on regulated deployments — synthetic data generation, supervised fine-tuning, DPO/RLHF, distillation, retrieval-augmented systems.
- Translate a customer decision problem into an evaluation contract that the model is held to in CI — and rebuild that eval as the workflow evolves.
- Ship models on-prem, in VPCs, or behind air-gapped boundaries; profile latency, calibrate confidence, and design the abstention behavior alongside product.
- Run the post-deployment loop: shadow-mode comparisons, calibration drift, retraining cadence, and the playbook for when production goes wrong.
- Set the modeling bar for the team we hire behind you — code review, eval discipline, internal tooling.
What we're looking for
- 5+ years building production ML systems, including at least one end-to-end deployment you would defend in a regulator meeting.
- Strong PyTorch fundamentals and direct experience fine-tuning open-weight LLMs (7B–70B class). You can discuss SFT, DPO, distillation, and quantization tradeoffs from first-hand work.
- A real evaluation muscle: held-out test sets, calibration, behavioral red-teaming, abstention. You have written eval harnesses, not just consumed them.
- Comfort working with non-AI engineers and operators. The product is the decision, not the model.
Bonus
- Prior work in a regulated sector (financial services, health, public-sector, or similar).
- Experience with on-prem, air-gapped, or VPC inference deployments.
- Open-source contributions to model evaluation, alignment, or interpretability tooling.
- Comfort with multimodal models (vision-language, document understanding).
How we hire
- 01 Intro call 30 min with one of the founders. Mutual fit, your trajectory, what we are building.
- 02 Technical conversation 60 min discussing a real engagement we're working on — modeling tradeoffs, evals, deployment constraints. No leetcode.
- 03 Paid work trial 1–2 days, paid at market rate. A scoped, real piece of modeling work on a sanitized dataset. We pair on it and review together.
- 04 Founder-team conversation 90 min with both founders. Vision, expectations, your questions on the operating reality.
- 05 Offer Within 48 hours of the final round.
How to apply
Hit the button below and you'll land on a short form where you can upload your CV and add a few links to work you'd point us at — a paper, a system you shipped, a write-up. No cover letter required.
- 01 Click every link. Open each project, paper, repo, or write-up linked from your CV and confirm it resolves to the right page. Broken links and dead deploys are the most common reason a strong CV gets passed over.
- 02 Verify your contact details. Email, phone number, location, and the best handle to find you on (LinkedIn, GitHub, X, your own site) — make sure each one is current. We will use exactly what's on your CV to reach you.
- 03 Name your file properly.
FirstName-LastName-CV.pdfis enough. PDF only, under 10 MB.
We reply to every application within five working days, including the ones that aren't a fit.