Figma Machine Learning Engineer interview questions

6 Machine Learning Engineers have shared notes from their Figma interviews, with the questions they were asked in each round. That is 46% of the 13 Figma interview notes on Screna AI.

6 Machine Learning Engineer interview notes

Figma·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Figma ML Engineer interview had a coding round with a grid traversal problem. Pretty standard stuff but worth knowing cold if you're going in.
  • Given a 2D grid where each cell is either land or water, count the number of distinct islands, where an island is a group of land cells connected in four directions.

“Classic grid traversal.”

View Post

Figma·Machine Learning Engineer·Onsite - System Design / Architecture

Jun 2026
Figma ML Engineer system design round, one big open-ended question about building a natural-language-to-HTML generation service. Pretty deep technically, they wanted you to go end to end from model choice all the way through serving and evaluation. Left feeling like I covered the surface but probably undersold some of the harder tradeoffs.
  • Design a system that accepts a natural-language description of a webpage and generates the corresponding HTML (and optionally CSS and JS).

“This one sprawled in every direction.”

View Post

Figma·Machine Learning Engineer·Onsite - System Design / Architecture

Apr 2026
Figma ML engineer interview with a meaty system design problem centered on document state and undo functionality. The kind of question that sounds straightforward until you actually have to think through the trade-offs.
  • You're given a layer-based document model (think design tool) that supports add, delete, move, and property change operations. Implement an undo feature that reverts the most recent operation. Walk through how you'd record operations and compare inverse-operation logging against full document snapshots.

“I started with snapshots because it felt safe, and the interviewer let me talk for a bit before asking what happens when the document is large.”

View Post

Figma·Machine Learning Engineer·Onsite - System Design / Architecture

Apr 2026
ML system design round at Figma for an MLE role. The whole session was one big open-ended question about building a home feed recommendation system from scratch, covering everything from retrieval to serving. Dense but actually pretty interesting if you've done this kind of work before.
  • Design a home feed recommendation system for a large consumer app. Walk through how you'd frame the problem, generate candidates, rank them, and serve the results at scale.
  • How would you handle candidate generation at scale? What retrieval approaches would you consider and what are the tradeoffs?
  • How would you structure the ranking model, and how do you handle real-time versus batch features in that pipeline?
  • Once you have ranked results, how do you re-rank them before serving? What factors beyond the model score matter?
  • How would you set up the serving infrastructure and measure whether the system is actually working in production?

“This is basically the canonical ML system design question and I still fumbled parts of it.”

View Post

Figma·Machine Learning Engineer·Technical Phone Screen

Apr 2026
Figma ML engineer interview with a coding problem centered on document editor state management. The problem was more design-heavy than I expected for an ML role, which threw me off a bit.
  • You're building a layer editor (think Figma-like) that supports operations like add, delete, move, and property changes, plus an Undo command and a batch primitive that groups operations into a single atomic unit. Implement Redo: after one or more Undos, the user can re-apply the most recently undone operation or batch. A new mutating operation after an Undo should clear the redo stack. Batched operations must behave atomically, same as with Undo.

“Took me a second to even process the scope of this.”

View Post

Figma·Machine Learning Engineer·Onsite - System Design / Architecture

Mar 2026
Got a system design question for an ML Engineer role at Figma that was way more software-architecture-heavy than I expected. The whole thing centered on building undo/redo semantics for a layer editor, which felt more like a backend infra problem than anything ML-adjacent.
  • Design a batch operation primitive for a layer/document editor where multiple mutations can be grouped together and treated as a single atomic unit for undo. How do you structure the operation log, handle nested batches, and deal with partial failures within a batch?

“Spent the first few minutes just making sure I understood what 'atomic undo' actually meant here.”

View Post