For AI labs and model teams in New York
Research Engineer Recruiters for AI Labs and Startups in NYC
Research ideas only matter once they run. We find research engineers for New York City AI startups and labs who have owned those runs, from the first ablation to a model the team shipped.

At a glance
Research roles we recruit for
Some labs need research engineers who mostly build training infrastructure. Others want them proposing experiments too. We agree which first.
Roles we recruit for
Research engineers, research scientists and ML engineers on model teams
Stages
Seed to Series C
Where
New York City. In office, hybrid or remote
Who we look for
Engineers who have run training at scale and turned ideas into results a team relied on
Speed
Our last 3 hires averaged 17 days
How you pay
Retained, success-based or embedded. Terms agreed upfront
How we assess
Research engineers who make experiments run at scale
For research engineers, we focus on the engineering behind the research: the runs, the infrastructure and whether another team could reproduce the result.
01
Ran training at scale
The largest runs they owned, the hardware behind them and what went wrong halfway through. How they handled the failures matters more than the size.
02
Turns ideas into results
One idea taken from a paper or a hunch to a result the team built on. We ask them to walk through each step, including the experiments that failed.
03
Designs clean experiments
Ablations, baselines, evaluation. We ask how they knew a result was real and not noise.
04
Writes code others reuse
Research code tends to rot. Training code and tools that other researchers kept using are a strong signal.
Training run ownership
Research outcomes
Experiment design
Compute expectations
Research depth
Training infrastructure
Research code quality

Who runs your search
Liam, Founder of StellarSearch
Our team brings 20+ years of search experience across engineering, AI/ML, product and GTM.
LinkedIn ↗
Client feedback
Feedback from founders and the engineers we placed



I'm not really sure how he does it, but he continues to find us top tier talent that fits exactly what we're looking for."











I'm very grateful they reached out."

Would highly recommend."

They did a great job explaining roles, companies, and interview processes and was friendly + supportive throughout."

Where our hires come from
We've placed people from teams like these
We look past titles to the work: ex-founders, early employees and people doing their best work outside the day job.




How we work
How we run your search
From the first call to an accepted offer. What we handle, and the few places we need you.
Learn what you're building, what the hire owns in the first 90 days, and the comp, equity and location you can offer.
One call with the founders.
Map the market from day one. If the problem is sourcing, pay, location, process or the profile, we tell you early.
Agree any changes to the brief.
Approach people who aren't applying anywhere, and assess each one on evidence before they reach you.
Nothing. This part is ours.
Send each candidate with our write-up, then schedule and keep the process moving.
Run your interviews.
Advise on the offer, handle the back and forth, and close.
Make the final call.
If the search isn't working, we tell you fast. We use market data to recalibrate instead of sending weaker candidates for weeks.
Ways to work with us
Three ways to work with us
Pick the model that fits the hire. All terms are agreed upfront, with no hidden fees.
Best for a priority hire
Retained
A dedicated search for the one hire that can't wait. We commit time to your role from day one.
Dedicated search from the first call
Market map shared with you
Fee agreed upfront
Best for a single hire
Success-based
We run the search and you pay when the hire is made.
No fee until you hire
Same assessment on every candidate
Terms agreed before we start
Best for several roles
Embedded
We work as part of your team across several hires, for a monthly fee plus a reduced success fee.
Monthly fee plus reduced success fee
One hiring standard across roles
Works alongside your in-house team
Start a search
What will your research engineer work on?
Tell us what you're building. We'll come back with how we'd run the search.
Looking for a role? See open jobs
Research engineer or research scientist: who should we hire?
Hire a research scientist if you need someone to set the research agenda, and a research engineer if you need ideas turned into training runs and results. Small labs often need people who can do some of both, and at many labs the two titles overlap.
Do research engineers need a PhD?
No. Many strong research engineers came from software or ML engineering, and the training runs and results they have owned predict more than a degree. A PhD matters more for research scientist roles that set a lab's direction.
Can we hire strong research engineers in New York?
Yes, New York has research teams at big tech labs, a growing number of AI labs and PhD programs at NYU, Columbia and Cornell Tech. Quant firms add engineers with rigorous experiment habits, though fewer have trained large neural networks. The pool is smaller than the Bay Area's, so we agree early whether relocation is part of the search.
What do research engineers want to know before they talk to us?
Compute first: how many GPUs they can use and how quickly they get them. Then whether they can publish, and how their time splits between research and product work. We get your answers in the brief so candidates hear them in the first call.
When should we hire a research engineer instead of an ML engineer?
If your edge comes from training your own models, hire research engineers; if you build on existing models, start with ML or AI engineers. An ML engineer gets models into your product and keeps them running, while a research engineer makes new training methods and experiments run at scale before anyone knows they will work.
How can a smaller lab compete with big AI labs for research engineers?
Rarely on cash. Smaller labs win on scope: owning whole training runs, a direct say in research direction and compute they can count on. Tell us what you can promise on compute and publishing, and we lead with it when we approach candidates.
How quickly will we see candidates?
We deliver exceptional candidate submissions within 24 hours, and for well-calibrated searches we can have candidates in process within the first few days, without compromising on the bar. Our last 3 hires averaged 17 days from submission to accepted offer.
We already have an in-house recruiter. Why use you?
Because the people you want usually aren't applying. Most of our clients already have founders, an internal talent team or both recruiting. We focus on the people they aren't reaching: we map the market and approach candidates who meet your real bar.
How much of the process do you own?
As much or as little as you need. At the simplest level, we source, qualify and close candidates. When we're embedded, we also help define profiles, calibrate pay, design interview processes, manage offers and advise founders on the close.
What if the search isn't working?
We tell you, fast. Because we map the market from the start, we can usually see whether the issue is sourcing, pay, location, the interview process or an unrealistic profile. We bring that data back and recalibrate instead of sending weaker candidates for weeks.
How do you charge?
We work retained, success-based or embedded, and agree everything upfront with no hidden fees. For individual searches, we offer retained or success-based terms. Higher-volume clients usually work with us embedded, for a monthly fee plus a reduced success fee. Placements are backed by a 60-day placement guarantee.
Hiring your founding team?
Tell us what you're building. We'll find the experts who can build and ship it.
Start Hiring