PwC AI Engineer candidates report behavioral and project discussion, practical AI concepts, SQL/data-cleaning work, take-homes, and—in one account—high-scale coding and system design.
$140K
Avg. Base Comp
$202K
Avg. Total Comp
3 rounds
Typical Rounds
2-4 weeks
Process Length
PwC AI Engineer interviews reported here emphasize explaining practical work clearly, then applying that thinking to data and AI problems. One candidate described an HR conversation of about 30 minutes centered on behavioral, fit, collaboration, and STAR-style examples; that person also encountered a psychometric assessment after applying. The technical discussion covered prior projects, including why particular tools and technologies were selected, alongside basic GenAI, agentic AI, machine-learning concepts, and current AI developments. A case study was part of that candidate’s process.
Another candidate reported three rounds with SQL joins and aggregations, a data-cleaning case, a logic-focused question with no single right answer, and a small deadline-bound take-home. Their advice was to make the details work, rather than relying on polish alone. Separately, one account described considerably more systems-heavy exercises: finding top-K IP addresses from a high-throughput rolling stream under tight memory constraints, followed by a distributed notification-system design. Prepare to explain tradeoffs, not simply name tools.
The reports show different levels of technical depth, so applicants should be ready for both conversational project defense and hands-on data, coding, or design work. End-to-end timing was not reported in the current accounts.
Synthesized from 4 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Pwc process.
This interview focused heavily on scale and optimization. In the Coding Round, I was tasked with tracking the Top K most frequent IP addresses from a high-throughput stream (10k+ requests/sec) in a rolling 5-minute window. Because strict memory constraints ruled out a standard HashMap, I successfully optimized the solution by combining a Count-Min Sketch for fixed-memory frequency approximation with a Min-Heap of size K. In the System Design Round, I designed a distributed notification system handling 100 million daily alerts. To solve their core requirements, I used Apache Kafka with isolated topics to prioritize critical OTPs over marketing blasts, and implemented a Redis distributed cache to manage unique transaction tokens, effectively preventing duplicate messages during downstream provider timeouts.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Pwc
Write a query to return the two students with the closest test scores and the score difference
| Question | |
|---|---|
| Size of Joins | |
| Hurdles In Data Projects | |
| RAG Strict Source Control | |
| Sort Strings | |
| Data Pipelines and Aggregation | |
| Data Preparation for Imbalanced Data | |
| RAG Hallucinations | |
| Overfit Avoidance | |
| Data Cleaning Experiences | |
| Why Do You Want to Work With Us | |
| Your Strengths and Weaknesses | |
| Clustering Basketball Players | |
| Creating Companies Table | |
| Feedback Sentiment Analysis | |
| 2nd Highest Salary | |
| Rolling Bank Transactions | |
| Merge Sorted Lists | |
| Longest Streak Users | |
| Top 3 Users | |
| Raining in Seattle | |
| Maximum Profit | |
| Find the First Non-Repeating Character in a String | |
| Bagging vs Boosting | |
| Skyscanner Partner ETL | |
| Real-Time Transaction Streaming | |
| P-value to a Layman | |
| Normalize Grades | |
| Fair Coin | |
| Using R Squared |
Synthesized from candidate reports. Individual experiences may vary.
One candidate reported a psychometric assessment after applying, followed by an approximately 30-minute HR interview. Candidates report behavioral and fit questions about teamwork, collaboration, and past experience, with STAR-style answers and clear communication playing a visible role.
A candidate reported discussing personal projects and prior experience, including defending the tools and technologies chosen. The conversation covered basic GenAI, agentic AI, machine learning, and current AI developments; a case study accompanied this technical discussion.
One candidate reported SQL joins and aggregations, a data-cleaning case, and a logic question without one right answer. That account also included a small take-home built under a deadline, where candidates may need to demonstrate careful execution as well as reasoning.
One candidate described a coding task involving top-K frequent IP addresses in a high-throughput rolling stream with memory constraints, followed by design of a distributed notification system. This suggests some processes may probe optimization, prioritization, duplicate prevention, and architecture tradeoffs.
