Candidates describe Deloitte AI Engineer interviews as 2-4 rounds mixing GenAI project walkthroughs, RAG and transformer questions, light Python coding, deployment tooling and behavioral conversations.
$115K
Avg. Base Comp
$130K
Avg. Total Comp
2-4 rounds
Typical Rounds
Not reported
Process Length
Candidates describe Deloitte's AI Engineer interviews as project-driven GenAI conversations more than algorithm drills. Several reports describe a walkthrough of resume projects, and the technical questions that follow often center on RAG: one candidate was asked how indexing, retrieval and ranking work internally, another to design a robust RAG pipeline, and a third to explain a RAG-based sentiment analysis solution. Transformer and attention mechanics came up in more than one account, as did agents, agent memory and hallucination reduction.
Coding appears to be light. Reports mention simple Python and OOP exercises, list comprehensions and a Fibonacci-style problem, while another candidate described a largely theoretical process with almost no coding. One candidate who did not advance after the second round said the questions turned to Docker, Kubernetes, AWS SageMaker and GitHub, and that they lacked deployment experience.
Experiences differ by interviewer and team. One candidate reported mostly data architecture questions (Kimball, Medallion, cloud versus on-premises). The one offer report involved four short rounds ending in a director-level behavioral conversation about team conflict and an HR call, and two accounts mention a final HR round. Most reports ended without an offer, so prepare to explain your projects clearly, to cover RAG and deployment, and to give concrete behavioral examples.
Synthesized from 7 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 Deloitte process.
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 Deloitte
Select the 2nd highest salary in the engineering department
| Question | |
|---|---|
| Merge Sorted Lists | |
| Size of Joins | |
| Raining in Seattle | |
| Maximum Profit | |
| Bagging vs Boosting | |
| Hurdles In Data Projects | |
| Real-Time Transaction Streaming | |
| Normalize Grades | |
| Using R Squared | |
| Cyclic Detection | |
| Skyscanner Partner ETL | |
| Missing Housing Data | |
| Precision and Recall | |
| Data Pipelines and Aggregation | |
| Find Duplicate Numbers in a List | |
| Data Preparation for Imbalanced Data | |
| Assumptions of Linear Regression | |
| Portfolio Platform Architecture | |
| Duplicate Rows | |
| String Palindromes | |
| FAQ Matching | |
| Classification and Regression | |
| Swap Variables | |
| Yelp-like System | |
| User Event Data Pipeline | |
| Bias vs. Variance Tradeoff | |
| Subway Machine Learning Model | |
| Seller Type Modeling | |
| Multicollinearity in Regression |
Synthesized from candidate reports. Individual experiences may vary.
One candidate described a brief GenAI screen of under 15 minutes about background and interest in generative AI. Not every account mentions a separate screen, so treat this as a possible but not universal opening.
Several candidates describe a technical round built around resume projects and frameworks they have used. Reported questions include handling OOV, self-attention, MCP, RAG pipeline design and basic Python or OOP exercises such as Fibonacci. One candidate described an almost entirely theoretical process with very little coding, so expect variation by interviewer.
Candidates report that a later technical round may go deeper: how RAG indexing and ranking work, fine-tuning hosted LLMs, vector databases, hallucinations in agents, and deployment with Docker, Kubernetes, SageMaker and GitHub. One candidate instead reported data architecture case studies, so confirm the team's focus with your recruiter if possible.
Some candidates report a managerial or director conversation. One described a fully behavioral director round about handling team conflict, and another a managerial round on transformer attention at a word level. Prepare STAR-style examples and a clear explanation of core model concepts, since the depth of the questions may vary.
Two candidates mention a final HR round. One described it as under 15 minutes and said the HR team sent regular updates between rounds. Another called it a formality. Outcomes varied: one offer was reported, while the other accounts ended without an offer.
