RepostedYahoo logo

Yahoo · Phone screen → Onsite

In process

Yahoo Mail Senior MLE Interview Experience: 1 Experience

Machine learning engineer · Full-timeDoctorateReported Nov 16, 2026
Shared byjanebeingjane

A 2026 Yahoo Mail senior MLE loop: a phone screen and a five-round panel with ML coding, recommendation design, AI debugging and product questions.

At a glance

  • Process: phone screen → one-day panel of five rounds: project deep dive, product manager, ML system design, AI coding, hiring manager (report 01)
  • Asked recently: logistic regression by hand, the Transformer architecture, a video recommendation design, five product-manager behavioral questions (report 01)
  • Difficulty: the candidate rated it easy and found the questions standard
  • Practise first: Design a Personalized Recommendation System, Logistic Regression with Gradient Descent (closest practice)

Coverage: One experienced machine learning engineer with a PhD, changing jobs, who came in through a recruiter for a senior role on the Yahoo Mail team. The report covers a phone screen and a five-round panel. It gives October to December 2026 as the interview period and went up on November 16, 2026.

What came up

The loop mixed ML coding, a recommendation system design, an AI-assisted debugging round, a product-manager round full of behavioral questions, and a hiring manager round. The candidate rated the experience very positively, but learned only in the last round that the team's work is mostly engineering rather than model training.

Topic Reports Last seen
ML coding: logistic regression from scratch with NumPy 1 of 1 Nov 2026 (report date)
ML fundamentals: the Transformer architecture, drawn on a whiteboard 1 of 1 Nov 2026 (report date)
ML system design: a video recommendation system 1 of 1 Nov 2026 (report date)
AI-assisted debugging of a multi-file mailbox report project 1 of 1 Nov 2026 (report date)
Product manager round: scoping, launch bars, trade-offs and AI use 1 of 1 Nov 2026 (report date)

Questions we found

Algorithms / Data Structures

Question Reports Last seen
Implement binary logistic regression with NumPy (closest practice) 1 Nov 2026 (report date)

System Design

Question Reports Last seen
Design a recommendation system for a video site (practice) 1 Nov 2026 (report date)

Behavioral (BQ)

Question Reports Last seen
A short intro and why you applied (practice) 1 Nov 2026 (report date)
Turning a vague product request into a scope, success criteria and a shared plan (practice) 1 Nov 2026 (report date)
Accepting a weaker model to save time or money, and who decided (practice) 1 Nov 2026 (report date)
A model that looks better offline but hurts the live metric just before launch (practice) 1 Nov 2026 (report date)
How AI tools fit into daily work, and building agents (practice) 1 Nov 2026 (report date)

ML knowledge

Question Reports Last seen
Explain the Transformer architecture and draw it on a whiteboard 1 Nov 2026 (report date)

AI Coding

Question Reports Last seen
Use AI tools to fix bugs in a multi-file project that builds a per-user mailbox report from a detailed README (the README and code aren't described) 1 Nov 2026 (report date)

1 detailed candidate timeline, loops, and screens

01: Senior MLE (PhD), phone screen and five-round panel

Interview period: 2026 Q4 · Report date: November 16, 2026 · Role: Senior machine learning engineer · Degree listed: PhD · Channel: Recruiter · Candidate's rating: Positive, easy · Result: Not stated; no news since the panel when posted

Phone screen

  1. ML coding. Write binary logistic regression by hand with NumPy. FastPrep's closest practice version is Logistic Regression with Gradient Descent, which uses plain arrays and a fixed gradient-descent loop instead of NumPy.
  2. ML fundamentals. Explain the Transformer architecture, drawing it on the CoderPad whiteboard.

The invitation to the panel came about a week later. The panel itself can be done in one day.

Panel

  1. Project deep dive. The candidate presented a project that used a large model for classification. It involved distillation with two levels of teacher models, which was hard to explain in the time, and the interviewers asked for detailed metrics. The candidate's takeaway: a simpler project may land better.
  2. Product manager round. All behavioral. The candidate listed the questions: a short intro and the reason for applying; a project where product's request was vague and how they scoped it, whether they set success criteria, how they planned it and shared the plan; how the team agreed on the launch bar with product and partners outside ML, and who decided; a time they accepted a weaker model to save latency or money and how they explained it; what to do when a model looks better offline but live tests or another team's checks show the product metric slipping close to launch; and, briefly, how they use AI tools and build agents. The interviewer recommended demonstrating agents during the later AI coding round. Practise them as a short intro and why you applied, shaping a vague product request, accepting a weaker model to save latency or money, a model that looks better offline but slips before launch and how AI tools fit into your daily work.
  3. ML system design. Design a recommendation system for a video site. In a reply, the candidate says these questions were standard, especially the recommendation design, and a framework answer works. FastPrep's practice version is Design a Personalized Recommendation System.
  4. AI coding. A project with many files that builds a report for each user's mailbox. A detailed README says what the report must contain, and the code has bugs to fix with AI tools. The candidate could choose the model, and Claude Code was allowed.
  5. Hiring manager. Only here did the candidate learn the role is engineering: the team prepares data for a research group and deploys and scales its models, while another group picks and trains the models. Asked whether they preferred research or engineering, the candidate said they train models, and the round ended after about 20 minutes.

Result

In a later reply the candidate says they told the hiring manager they weren't interested in a pure engineering role, and they hadn't heard back since.

Keep practicing

Sources: Anywhere useful interview info can be found.

0 total reactions. Your reactions: none.
0 comments

Comments

0

No comments yet — be the first to share what you know.