UiPath Interview Questions — Page 1

UiPath·Software Engineer·Onsite - System Design / Architecture

Jun 2024

Virtual onsite for a Software Engineer role at UiPath, conducted by a principal engineer. The whole thing was 60 minutes and included a project deep dive with diagramming done directly in HackerRank, which was a bit of an awkward setup.

Walk through a past project in depth, including drawing out diagrams to illustrate the architecture or design.

“The diagramming part inside HackerRank was not something I expected.”

View Post

UiPath·Software Engineer·Onsite - Multi Round

Apr 2026

Three-round backend interview at UiPath for an SDE-2 role covering DSA, a live coding exercise, and low-level design. Finished with a rejection that felt pretty arbitrary given how close to complete everything was, and the last interviewer seemed checked out before the round even started.

Solve two medium-to-hard algorithmic problems covering dynamic programming, strings, and greedy approaches within one hour.
Build a working event booking system using a backend framework of your choice, including business logic, authentication, and schema design, within one hour.
Design a bill-splitting application similar to Splitwise, including schema design and expense handling logic, covering all stated requirements.

“Finished both in time which felt good.”

View Post

UiPath·Machine Learning Engineer·Technical Phone Screen

Apr 2026

Technical phone screen for an ML Engineer role at UiPath. One question, pretty deep on transformer internals, and the follow-ups were where things got interesting.

What is the time and memory complexity of self-attention for a sequence of length N with model dimension d?

“I got the O(N² · d) time part right away but fumbled explaining why memory is O(N²) specifically.”

View Post

UiPath·Machine Learning Engineer·Technical Phone Screen

Apr 2026

Interviewed for an MLE role at UiPath and got hit with a pretty thorough overfitting question that turned into a whole conversation about regularization tradeoffs. Felt like a technical phone screen vibe, not a casual chat.

Your model is overfitting. Walk me through the full toolkit you'd use to address it and explain when each approach is actually appropriate.

“This sounds like a warmup but it isn't.”

View Post

UiPath·Machine Learning Engineer·Technical Phone Screen

Apr 2026

UiPath ML Engineer interview with a case study focused on house price prediction. The whole thing was basically one big feature engineering discussion, which I wasn't expecting to go as deep as it did.

You're given a dataset with only two features, area (numeric) and neighborhood (categorical), to predict house prices. Walk through all the feature engineering and preprocessing you'd do before fitting any model.
For a high-cardinality categorical feature like neighborhood, what encoding strategies would you consider and what are the tradeoffs?
What model would you start with for this problem and why, and how would you choose your evaluation metric given the distribution of house prices?

“This started straightforward and then kept expanding.”

View Post

UiPath·Machine Learning Engineer·Technical Phone Screen

May 2026

Interviewed for an ML Engineer role at UiPath and got a coding question that felt more like a CS fundamentals refresher than anything ML-related. The focus on sorting algorithms with full complexity analysis was a bit unexpected.

Implement a bidirectional bubble sort (cocktail shaker sort) for an integer array. Optimize it with early termination and shrinking bounds, write basic tests, and analyze its time and space complexity, stability, and how it compares to standard bubble sort across best, average, and worst cases in terms of swaps and comparisons.

“This one took me a second to even remember what cocktail shaker sort was.”

View Post

UiPath·Machine Learning Engineer·Technical Phone Screen

May 2026

Whiteboard-heavy technical screen for an ML Engineer role at UiPath. The whole session was basically a math notation audit, write the formula, get it exactly right, move on. Unforgiving if you're sloppy with summation indices or mix up scalar and vector notation.

Write the dot product formula for two vectors u and v in R^n from memory, with correct notation.
Write the cosine similarity formula between two vectors u and v, including the correct denominator.
Write the Euclidean (L2) distance and Manhattan (L1) distance formulas between two points.
Write the Mean Squared Error formula for n predictions versus targets, then derive the partial derivatives with respect to both the weight and bias parameters.
Write the min-max normalization formula and the z-score standardization formula.
As a bonus: write the sigmoid function and its derivative, the softmax function and its derivative, and binary cross-entropy.

“Seemed trivial going in.”

View Post

UiPath·Software Engineer·Technical Phone Screen

May 2026

Coding round at UiPath for a software engineer position. The problem was a classic number conversion question, nothing too wild, but it's the kind of thing that trips you up if you haven't thought about the subtractive cases recently.

Given an integer in the range 1 to 3999, convert it to its Roman numeral representation, accounting for subtractive notation like IV, IX, XL, and so on.

“I knew the symbols but blanked on how to cleanly handle the subtractive pairs.”

View Post

UiPath·Machine Learning Engineer·Technical Phone Screen

May 2026

Technical interview at UiPath for an ML Engineer role, focused almost entirely on Transformer internals. One long question that kept branching into sub-questions. Felt like a grad school oral exam more than a standard interview.

Walk through the Transformer architecture in full detail, covering both encoder and decoder layers, including multi-head attention, residual connections, layer normalization, and the feed-forward network. How do Pre-LN and Post-LN differ?
Why does a Transformer need a position-wise feed-forward network after the attention layer? What can the FFN do that attention alone cannot?
Using a concrete configuration like d_model=512 and 8 heads, trace the exact matrix shapes through a full attention block: Q/K/V projections, attention scores, softmax, weighted sum, head concatenation, output projection, residual addition, layer norm, and both FFN linear layers.
Derive the computational complexity of self-attention with respect to sequence length n and model dimension d. Where does memory become the bottleneck?
What are the trade-offs between different implementation choices in Transformers: fused vs. separate Q/K/V projections, multi-query and grouped-query attention, positional encoding schemes like RoPE or ALiBi, and FFN variants like SwiGLU or RMSNorm?
How do encoder and decoder layers structurally differ, and how do token representations change as they pass through successive layers in the stack?

“This is where the interview started and basically never left.”

View Post

UiPath·Software Engineer·Onsite - Behavioral / Leadership

May 2026

Behavioral round at UiPath for a software engineer role, focused pretty heavily on past project ownership. One big question that took up most of the time.

Walk me through a project you worked on end-to-end: what problem it solved, what you personally owned, the technical and product decisions you made, whether you led it or supported someone else, what trade-offs you faced, and what you'd do differently now.

“This is the kind of question that sounds easy until you're actually in it and realize you've been rambling about context for three minutes without getting to the interesting part.”

View Post