Jane Street Interview Questions — Page 5

Jane Street·Machine Learning Engineer·Onsite - System Design / Architecture

Jun 2026

Jane Street ML Engineer interview, system design round focused entirely on building a production-grade time-series forecasting platform from scratch. One very long question that basically covers everything you'd need to know about MLOps. Felt like a PhD thesis defense more than a typical interview.

Design a complete end-to-end training framework in PyTorch for time-series forecasting, covering data ingestion with windowed sampling and feature generation, config management and reproducibility, training loop with mixed precision and gradient clipping, early stopping and checkpointing, hyperparameter tuning, experiment tracking, model registry with promotion gates, batch and streaming inference with latency SLOs, feature parity between train and serve, monitoring with drift detection and automated retraining triggers, and a full testing and CI/CD strategy. Also describe the high-level module diagram and key interfaces between components.

“This was basically the whole interview.”

View Post

Jane Street·Machine Learning Engineer·Technical Phone Screen

Jun 2026

Coding round for an MLE role at Jane Street. One question, but it was a lot more implementation-heavy than I expected from a finance firm.

Build a simplified code editor (think VSCode-style) that supports collapsing and expanding code blocks. For example, given a 5-line snippet where lines 2 through 4 form a for-loop block, calling shrink(2) should collapse that block down to just its header line, resulting in a 3-line view.

“I spent the first few minutes just trying to figure out the right data structure.”

View Post

Jane Street·Software Engineer·Technical Phone Screen

Jun 2026

Jane Street SWE interview with a pretty gnarly streaming systems question. The core problem was interesting but the follow-up variant is what really tested whether you'd actually thought through the edge cases.

You have a stream transformer that reconstructs a sparse code-time matrix. Input is batches of (timestamp, code, value) tuples and output is one (timestamp, row) pair per timestamp, with missing entries filled as -1. Now suppose the tuples within a batch are no longer sorted by code, only timestamps are guaranteed non-decreasing across batches. How do you modify the transformer to still emit correct rows, and what are the time and space tradeoffs?

“This is the kind of question where the base case feels manageable and then the variant exposes whether you actually understand the data flow.”

View Post

Jane Street·Software Engineer·Technical Phone Screen

Jun 2026

Jane Street SWE interview, coding round focused on a trading systems problem. Pretty domain-flavored but the core was just a sorting exercise with a twist at the end. Left feeling okay about it but the follow-up question made me second-guess myself.

Given an array of trade execution records that may arrive out of order, implement a function that sorts them into a canonical ordering: first by timestamp ascending, then by symbol lexicographically, then by trade ID lexicographically as a final tiebreaker.
How would you modify your approach if trade IDs are no longer guaranteed to be unique, meaning duplicate or replayed records could appear in the feed?

“Felt straightforward at first and it basically was.”

View Post

Jane Street·Machine Learning Engineer·Onsite - System Design / Architecture

Jun 2026

Jane Street ML Engineer interview with a pretty gnarly systems design question. Not the typical ML theory stuff I prepped for, which threw me off a bit. The question was open-ended enough that I kept second-guessing my own design choices throughout.

Design and implement a compiler or interpreter for a simple custom programming language. Walk through your object-oriented design, the compilation pipeline stages, and how you'd approach debugging and fixing errors in the system.

“This one took me a minute to even figure out where to start.”

View Post

Jane Street·Software Engineer·Technical Phone Screen

Jun 2026

Jane Street SWE interview with a two-part coding problem centered on building a collapsible code editor feature. The problem was more involved than it looked at first glance, especially the nesting independence requirement in part two.

Given a list of lines representing source code where each line has at most one brace as its last character, write a function that returns a mapping from each opening brace's line index to its matching closing brace's line index, in a single pass.
Extend the brace-matching solution with a toggle and getText interface that supports collapsing and expanding individual brace pairs, where each pair's collapsed state is fully independent regardless of nesting.

“Stack-based solution, pretty standard once you see it.”

View Post

Jane Street·Software Engineer·Onsite - System Design / Architecture

Jul 2026

Jane Street software engineering interview where I built a two-player piece-dropping game engine from scratch and then got asked how I'd productionize it as a library or API. The design follow-up was the real interview.

You've just coded a working game engine under time pressure as a single script. If you had to turn this into a library other engineers depend on, or expose it as a network API, what would you change and what would you add?
How would you add a new rule variant, like diagonal wins or a third player, without breaking existing users of the library?
If this became a network API, would you make the service stateful or stateless? What are the trade-offs?
Once other teams are depending on the library, how do you roll out a breaking change to the public interface?
If you only had time to write five tests before shipping, which behaviors would you pin down first?

“This is where I fumbled a bit.”

View Post

Jane Street·Machine Learning Engineer·Technical Phone Screen

Jul 2026

Jane Street ML Engineer interview, one big technical question that basically covered everything you'd ever want to know about time series forecasting. It was a lot to hold in your head at once.

You're given a time-series dataset with timestamps and numeric targets. Walk through your full pipeline: preprocessing, modeling choices, evaluation strategy, and how you'd handle concept drift over time.

“This question is basically five questions glued together and they let you run with it.”

View Post

Jane Street·Data Scientist·Technical Phone Screen

Jul 2026

Two pretty intense quant/algo questions for a Data Scientist role at Jane Street. Both required actual derivations and working code, not just hand-waving. The kind of interview where you either know the math or you don't.

You're at position k in a 100-person queue when a second checkout counter opens. Each other customer independently moves to the new queue with probability 1/2, and if you also move, your spot among the movers is uniformly random. Both counters serve at the same rate. Derive your expected wait time for staying vs. switching, and find the threshold k where switching becomes the better choice.
Given an integer n in cents and coin denominations of 1, 5, 10, 25, and 100 cents, count the number of distinct combinations (order doesn't matter) that sum to exactly n. Formalize it, design an algorithm for n up to 10,000, and analyze time and space complexity.

“This one took me a while to set up properly.”

View Post

Jane Street·Data Scientist·Technical Phone Screen

Jul 2026

Jane Street Data Scientist interview with a probability/game theory problem involving optimal dice strategy. The math is clean but the reasoning chain to get there is genuinely tricky, especially under pressure.

Alice rolls a fair 30-sided die and Bob rolls a fair 20-sided die. Whoever shows the higher number wins, with ties going to Bob. Bob can reroll once after seeing his first result, but he never sees Alice's number before deciding, and his second roll is binding. Assuming Bob plays optimally, what is the probability that Alice wins?

“This one took me a while to set up properly.”

View Post