picked.ai/hire/machine-learning-engineer
Hire a machine learning engineer
How to hire a
machine learning engineer.
Salary bands, time-to-fill, what good actually looks like, and the assessment we put every candidate through. Built on nine years of validated psychometric data from Neuroworx.
Post this role
~30 seconds
Title, level, three must-haves. We syndicate, screen, score, and shortlist. First 50 candidates free.
Post this role for free →
£140,000
Median salary · United Kingdom
£115-£165k senior range
42d
Typical time-to-fill
down to 4.8d with Picked
196
Applicants per role
Series A · public posting
9 / 196
Make it to interview
industry · pre-AI screening
Salary bands
What it costs to hire one.
2026 H1 · base + variable
LevelUSUKEURemote
IC1 / Junior$109-$149k£62-£85k€66-€91k$82-$112k
IC2 / ML engineer$149-$207k£85-£118k€91-€126k$112-$156k
IC3 / Senior ML$201-$289k£115-£165k€123-€177k$152-$218k
IC4 / Staff ML$271-$382k£155-£218k€166-€233k$205-$288k
IC5 / Principal ML$368-$516k£210-£295k€225-€316k$277-$389k
Source · 2,847 anonymised offers placed via Picked Q4 2025 to Q1 2026. Updated monthly.
What good looks like
Past credentials. Present signals.
01
Modelling judgement
Knows when a logistic regression beats a transformer. Has chosen the smaller model in the last six months.
02
Production discipline
Has shipped a model with monitoring, a rollback path, and a documented owner. Not a notebook in production.
03
Evaluation honesty
Reports the baseline alongside the model. Will tell you when the lift is inside the noise.
04
Cost awareness
Knows what a training run costs and what serving costs per thousand requests. Has killed a job to save money.
05
Comms with researchers
Can read a paper and ask the one question that exposes whether the result generalises.
Nine years of Neuroworx outcome data on 14,000 machine learning engineer hires says: the candidates rated high on modelling judgement + evaluation honesty outperform high-prestige-CV candidates by a wide margin in the first twelve months.
Read the paper →
What Picked puts every candidate through
Four stages. Each one written down.
From the Neuroworx item bank · 1,247 items live
01
Triage
CV + cover screened against must-haves. Title, level, language, region, three skills you set. Hard filters first. The model spends no time on people you would not even read.
~12 seconds · $0.12
02
Conversational screen
A 12-minute voice or chat conversation against six structured probes: motivation, scope of last role, the thing they are proudest of, biggest production mistake, what they would build with a free month, why this role.
~12 min · $0.31
03
Role-fit assessment
A scoped task drawn from the bank. Submit a short writeup. Scored against a published rubric on judgement, craft, and pragmatism.
30-60 min · $0.28
04
AI first-round interview
A 20-minute voice interview against six items drawn from the IC3 item bank. Includes behavioural ("tell me about a time"), technical or scenario ("walk me through how you would approach"), and an ownership probe.
~20 min · $0.24
The 1,247 items for machine learning engineer are split across 42 sub-rubrics. Every item has been calibrated against twelve-month performance outcomes from 14,083 prior hires.
See sample items ↗
Sample interview questions
A few from the item bank.
01
Tell me about the last model you shipped to production. What degraded first?
Listen for: A named model, a real input distribution shift, the signal they caught it on, the action they took.
02
When did you last decide not to use ML for something?
Listen for: A specific problem, the heuristic or rule they shipped instead, the reason ML was a worse fit.
03
Walk me through your last training run. What was the baseline?
Listen for: A real baseline (often a simple model or a rule). The honest delta. The reason the delta is worth the complexity.
04
Describe a time your model gave the right answer for the wrong reason.
Listen for: A leakage story, a spurious feature, a confounder. The way they found it.
05
How do you decide a model is ready for production?
Listen for: A real bar. Offline metrics, online shadow run, rollout plan, monitoring in place. Not a wishlist.
See all 30 questions →
Job description template
Copy. Paste. Replace the bracketed bits.
machine-learning-engineer.md
206 words · 3 min read
# [Senior/Mid/Junior] machine learning engineer · [city] or remote

We are hiring our [Nth] ML engineer. You will own a model in production end-to-end (currently: [the X model, ~Y predictions/day]). The team is [N] ML engineers, [N] researchers, and [N] platform engineer. [How you evaluate]. [How you ship].

## You will be good here if

- You have shipped a model to production and watched what happened next.
- You write the baseline before you write the model.
- [The third trait you actually care about].

## What we use

[Frameworks, infra, serving stack]. We will not retrain you on a stack, but we do not care if you came from [adjacent stack].

## Compensation

[Currency][low] to [high] base. [low] to [high] percent equity. We post the band because we mean it.

## How we hire

30-second post, screen + assessment via Picked, 20-min interview via Picked, one on-site half-day with the team. We aim to give a yes or no in 7 days.

## What is the rubric

Modelling judgement, production discipline, evaluation honesty, cost awareness, comms with researchers. Not credentials.

That is the JD. Apply via [link] or just send a paragraph about the last model you shipped to production.

[Your name], [your title]
Full JD template page →
Hire a machine learning engineer.
By Friday.
Post this role for freeSee a real dashboard →
$0.99 per AI-vetted candidate. First 50 free.
Hire machine learning engineers · Picked.ai