Lyft interview notes, page 10 of 10

Large Enterprises

Most of these Lyft write-ups come from Software Engineers (41), then Product Managers (28) and Data Scientists (19). The most reported rounds are the technical phone screen (42), the product sense round (29), the system design round (12) and the behavioral / leadership round (4).

97 interview notes · updated Sep 2026

Lyft·Software Engineer·Technical Phone Screen

Jul 2026

Lyft software engineer interview with a meaty coding problem centered on building an in-memory key-value store from scratch. The problem had more moving parts than I expected and I spent way too long second-guessing the transaction design.

Design and implement an in-memory key-value database that supports SET, GET, DELETE, COUNT, BEGIN, ROLLBACK, and COMMIT commands, with average O(1) per operation and correct transactional semantics including nested transactions.

“The COUNT part is what tripped me up first.”

View Post

Lyft·Data Scientist·Technical Phone Screen

Jul 2026

Lyft data science interview with a probability/stats question built around a pricing experiment scenario. The setup was clever and required more careful modeling than it first appeared.

In a pricing experiment, each new rider is offered two rides on their first day, where each ride is priced reasonably with some independent probability P. What is the probability that the rider takes exactly one ride the following day? What is the probability they become a permanent rider who rides every day going forward?

“I stared at this for a second because the 'permanent rider' part felt open-ended.”

View Post

Lyft·Machine Learning Engineer·Technical Phone Screen

Jul 2026

Interviewed for an ML engineering internship at Lyft, two conceptual questions back to back, both on the heavier theory side. Felt like they wanted to see if you actually understood the underlying mechanics or were just pattern-matching buzzwords.

What is the difference between a transformer model and an embedding model? What does each one produce, how are they typically trained, and when would you choose one over the other?
What is the difference between pretraining and fine-tuning? Why is pretraining valuable, what actually changes during fine-tuning, and what failure modes or trade-offs should you be aware of?

“I started okay, said transformers produce token-level outputs across a sequence while embedding models collapse everything into a single fixed vector.”

View Post

Lyft·Software Engineer·Technical Phone Screen

Jul 2026

Lyft coding round with two meaty problems back to back. Both were more involved than typical leetcode fare and required you to actually think through data structures and tradeoffs rather than just pattern-match to a solution.

Implement a cursor-based pagination function get_page(user_id, page_size, cursor) over an in-memory collection of transaction records. Records have txn_id, user_id, amount, and created_at fields. Results should be ordered by created_at descending with txn_id ascending as a tiebreaker. The cursor must be opaque and guarantee stable pages even when new transactions are inserted mid-session. Discuss your data structures, how you encode and decode the cursor, and the time and space complexity.
Design and implement a time-versioned key-value store supporting set(key, value, t) and get(key, t), where get returns the value with the largest timestamp less than or equal to t, or null if none exists. Optimize for heavy read workloads (millions of queries). Achieve O(log n) per operation per key, handle duplicate timestamps deterministically, and discuss memory tradeoffs and concurrency safety for readers.

“The stable-page constraint is what makes this hard.”

View Post

Lyft·Data Scientist·Technical Phone Screen

Jul 2026

Lyft DS interview with a meaty case study spanning root cause analysis, pricing strategy, and experiment design all rolled into one question. It's the kind of thing that sounds manageable until you're 20 minutes in and realize you've barely scratched the surface.

Ride wait times at Lyft have increased 20% month-over-month. How would you structure a root cause analysis to systematically identify why, including defining KPIs and slicing by geography, time, and supply/demand dynamics?
How would you assess the costs and benefits of switching from static to dynamic pricing at Lyft, quantifying trade-offs across wait times, revenue, rider experience, and driver earnings?
Design an experiment to test whether dynamic pricing reduces wait times without hurting other key metrics, accounting for interference in a two-sided marketplace.

“I started with the obvious cuts: city, time of day, ride type.”

View Post

Lyft·Software Engineer·Onsite - System Design / Architecture

Jul 2026

Lyft system design round focused entirely on building a news feed backend. It was one of the more thorough design interviews I've done, covering basically every layer of the stack from data modeling to failure handling.

Design the backend for a social app's news feed. Walk through requirements, data model, APIs, and your overall architecture.
Compare fan-out-on-write versus fan-out-on-read for a feed system. Which would you pick given high write and read loads, and why?
How would you handle ranking and real-time updates in the feed?
How do you handle deduplication, privacy controls, and user blocks in the feed pipeline?
Walk through your caching strategy, storage choices, and how you'd shard the data.
How would you handle cold-start for new users and backfilling feeds after someone follows a new account?
What failure scenarios would you plan for, and how would you monitor the feed system in production?

“I started with requirements which felt fine, but I spent too long on the functional side and had to rush the non-functional stuff.”

View Post

Lyft·Software Engineer·Onsite - Multi Round

Sep 2026

Four rounds at Lyft covering CS fundamentals, a laptop coding problem, an experience/behavioral round, and system design. The coding rounds went smoothly but the experience round ran over time and left me a bit uneasy about how it was received.

Given an array of integers, find the missing number in a sequence.
You have a function that checks whether a given version is the first bad one in a series. How do you find it efficiently?
Design an in-memory key-value store that can be distributed across multiple nodes.
Walk me through your past projects and relevant work experience.
Design a data management system, covering overall architecture and key design decisions.

“Finished this faster than expected and had time to write extra test cases and talk through what prod readiness would look like.”

View Post