Turing is contracting experienced ML engineers to build and solve MLE Bench–style evaluation tasks on production-grade machine learning codebases. The work is hands-on Python: training, evaluation and inference pipelines, dataset and metric preparation, and debugging realistic ML systems so frontier labs can measure how well AI agents handle real engineering work.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work actually is

MLE Bench evaluations test whether an AI system can do the job of a machine learning engineer end to end — read an unfamiliar repo, prepare data, train a model, evaluate it honestly, and ship something that runs. Your role is to sit on the other side of that: constructing the tasks, building reference solutions, and judging model attempts. Day to day that means working inside real-world ML codebases, building and modifying training/evaluation/inference pipelines, preparing datasets, features and metrics, and debugging production-like systems for correctness and performance. A meaningful slice of the time goes to analysing model behaviour — where an attempt silently leaked test data, trained on the wrong split, optimised a metric that doesn't answer the question, or produced a number that looks great and is wrong.

Code review is part of the loop, not an afterthought. Tasks and solutions are reviewed by other engineers and researchers, and your own submissions will be read closely for reproducibility, clarity, and whether someone else can rerun them and get the same numbers.

What the screen looks for

The stated evaluation is a single 60-minute technical interview with a live coding challenge. Expect to write real Python under observation and to be asked follow-ups about choices you made — why that metric, why that split, what happens when the data is imbalanced or the pipeline is slow. Turing asks for a minimum of 3+ years as an ML engineer or ML-focused software engineer, strong Python for ML and data workflows, direct experience with PyTorch, TensorFlow, JAX or similar, and the ability to navigate a large codebase you did not write. Clear spoken and written English matters because task specifications and failure-mode write-ups are the deliverable as much as the code.

Logistics

Fully remote contractor assignment, no medical or paid leave.
Minimum 4 hours per day and 20 hours per week, with 4 hours overlapping PST.
Initial contract of 3 months, adjustable based on engagement.
Hiring is restricted to India, Pakistan, Nigeria, Kenya, Egypt, Ghana, Bangladesh, Turkey, Brazil and Mexico.
Pay is not disclosed in the listing; rates on Turing's ML engineering pools are typically negotiated per engagement and by seniority. Treat any figure you hear secondhand as unverified until it is in your contract.

MLE Bench – ML Engineers

What the work actually is

What the screen looks for

Logistics

Interview prep — 10 questions