Uber MLOps Engineer Interview Guide

The short answer: Uber does not use the title MLOps Engineer. Its ML platform work is posted as Software Engineer roles on Michelangelo, Uber's in-house ML platform, for teams such as the Feature Store and AI Infra Training. Uber's ML & AI Engineering interview guide lists a 30-minute recruiter call, a 60-minute technical phone screen, then 3 to 6 interviews of about 60 minutes depending on level, covering coding, design and architecture, and collaboration. Uber does not say whether ML platform candidates get this guide or the Backend Engineering guide, so confirm which one applies with your recruiter. AI tools are not allowed unless your recruiter says otherwise. Last reviewed October 2026.

Uber MLOps engineer interview: quick answers

The Uber MLOps engineer role

What Uber MLOps engineers work on, the skills its job postings ask for, and what a week can look like. Uber posts the role as Sr Software Engineer - ML platform, Senior Software Engineer - ML Infra and Staff Software Engineer - ML Infra, with body titles such as Senior Software Engineer - AI Platform (Michelangelo Feature Store) and Sr Software Engineer - AI Infra - Training. It has no MLOps Engineer title.

Uber MLOps engineer levels and pay

Your level sets the bar you are interviewed against and the pay band of the offer.

Uber's interview guides use levels L3, L4, L5A, L5B and L6+, while postings say Senior or Staff, and Uber does not say which guide level each title maps to. No L3 or L4 ML platform posting was live, so that row has no pay.

Pay in Uber's US job postings

Uber's postings list a base salary range by location and say the role is eligible for a bonus and may include equity. The ranges were the same in every listed city.

The Staff Software Engineer - Engineer posting is a hybrid role that also asks for years of building ML models. No L3 or L4 ML platform posting was live, so there is no junior range. Ranges seen October 29, 2026.

Uber MLOps engineer interview process

Every step below comes from Uber's ML & AI Engineering interview guide, which lists a different loop for each level (L3, L4, L5A, L5B and L6+). Uber does not publish what the specialization round asks an ML platform candidate, and we found no readable candidate report on an ML platform loop.

Based on Uber's own pages.

Uber says interviews typically last 60 minutes, recruiters usually follow up within 2 to 3 business days after each one, and rescheduling is limited. It publishes no total timeline. Last reviewed October 2026.

What is different at Uber

• The title says Software Engineer. Uber files ML platform work as Software Engineering jobs, sometimes under a generic listing title such as Staff Software Engineer - Engineer, with the real title (for example Staff Software Engineer - ML Infra) in the body. Machine Learning Engineers on product teams such as Maps build models and use Michelangelo as a tool, and scientists go through a separate Sciences loop.
• You would work on Michelangelo, Uber's in-house ML platform since 2016. Uber reported about 400 active ML projects, more than 20,000 training jobs a month, more than 5,000 models in production and 10 million real-time predictions per second at peak in May 2024. Named parts include the Palette feature store, the Gallery model registry, Triton serving, Ray for training and a Gen AI Gateway for LLMs.
• GPU infrastructure is a large part of the job. Uber moved all ML projects to Ray on Kubernetes at the start of 2024, with a global control plane that routes jobs to resource pools by ownership and load, and reported training 1.5x to 4x faster. Its AI infrastructure team also benchmarked 17 GPU and CPU SKUs in 2023.
• Which guide applies is not stated. Uber's engineering hub says the process is consistent across engineering roles with different technical focus. The Backend Engineering guide adds the online coding assessment before the phone screen at L3, and the ML & AI Engineering guide does not, so ask your recruiter.
• The loop grows with level. L3 has three interviews after the phone screen, L6+ has two screens and six interviews, and the previously solved design round only appears from L5A.
• The AI rule is strict by default. AI tools, including code auto-completion, are not allowed unless your recruiter says otherwise, and some technical interviews may provide an AI agent that the recruiter tells you about ahead of time.

Uber MLOps engineer interview questions

Questions from our bank, then practice prompts we wrote for this guide around Uber's work.

1from typing import Dict, List, Tuple
2
3def route_jobs(pools: List[Tuple[str, str, int]],
4               jobs: List[Tuple[str, str, int]]) -> Tuple[Dict[str, str], List[str]]:
5    """Assign each job to a pool by ownership, then load, or queue it."""
6    pass

1from typing import Dict, List, Tuple
2
3def route_jobs(pools: List[Tuple[str, str, int]],
4               jobs: List[Tuple[str, str, int]]) -> Tuple[Dict[str, str], List[str]]:
5    free = {name: cap for name, _, cap in pools}
6    owners = {name: owner for name, owner, _ in pools}
7    assignments, queued = {}, []
8    for job_id, owner, gpus in jobs:
9        chosen = None
10        for group in (owner, 'shared'):
11            fits = [n for n in free if owners[n] == group and free[n] >= gpus]
12            if fits:
13                chosen = min(fits, key=lambda n: (-free[n], n))
14                break
15        if chosen is None:
16            queued.append(job_id)
17        else:
18            free[chosen] -= gpus
19            assignments[job_id] = chosen
20    return assignments, queued
21
22# Complexity: O(j x p) time for j jobs and p pools, O(p) space.

How to prepare

A plan for the Uber loop, most important first:

1. Ask your recruiter which guide applies (ML & AI Engineering or Backend Engineering), your target level and what the specialization round covers.
2. Practice algorithms and data structures for two 60-minute coding rounds, using the problem types in Uber's examples: arrays, strings, two pointers, dynamic programming and search.
3. Practice ML platform design with real numbers: real-time feature serving, GPU job scheduling across clusters, model serving and an LLM gateway.
4. Prepare a deep dive on a system you built, with its scale, failures and trade-offs, for the previously solved design round from L5A up.
5. Read Uber's Michelangelo and Ray on Kubernetes blog posts so you can talk about the platform you would work on.
6. Write STAR stories for Collaboration & Leadership mapped to Uber's values and success attributes, and practice without AI help.

Practice path

Courses that cover the rounds above:

• Model Production
• System Design
• ML System Design
• Data Structures and Algorithms
• GPU Programming
• Data Pipelines and Orchestration