
One candidate reports a two-round technical Accenture ML Engineer interview: a project walkthrough and F1-from-scratch coding, then SVM, KNN, NLP and RAG fundamentals. Scheduling was slow and the outcome was no offer.
$138K
Avg. Base Comp
$170K
Avg. Total Comp
Not reported
Typical Rounds
Not reported
Process Length
The Accenture ML Engineer interview described here was technical in both rounds, and it mixed practical coding with conceptual questions. The candidate, who applied through LinkedIn and did not receive an offer, began with a detailed walkthrough of one of their own ML projects. The interviewer asked about data preprocessing, hyperparameter tuning and the choice of evaluation metrics, so be ready to defend each decision you made rather than just list the steps.
The coding task the interviewer seemed most focused on was writing an F1 score calculation from scratch, without sklearn. Practice computing precision, recall and F1 by hand from raw predictions and labels.
The second round, which came about a month later, moved to fundamentals: support vectors in SVM, how KNN works and whether it is trained, deep learning versus machine learning, and CNN architecture. It then turned to NLP and generative AI, covering stemming versus lemmatization and when to prefer each, plus practical RAG questions. One question asked why you would use an LLM in a RAG system if you could simply fetch documents from a vector database. Prepare a clear answer about what generation adds over retrieval.
Timing was the main frustration. The candidate waited about two weeks after several recruiter calls, and a connectivity problem on the interviewer's side pushed the first session back two days. The report is a single experience, so use it as a starting point for preparation and not as a fixed format.
Synthesized from 1 candidate report by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Accenture 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 Accenture
Given two sorted lists, write a function to merge them into one sorted list.
| Question | |
|---|---|
| Raining in Seattle | |
| Assumptions of Linear Regression | |
| Bagging vs Boosting | |
| Bias vs. Variance Tradeoff | |
| Encoding Categorical Features | |
| Bank Fraud Model | |
| Missing Housing Data | |
| Payment Data Pipeline | |
| Data Pipelines and Aggregation | |
| Hurdles In Data Projects | |
| Different Parcel Effectiveness | |
| Xgboost vs Random Forest | |
| Justify a Neural Network | |
| Confidence Interval Explanation | |
| Data Preparation for Imbalanced Data | |
| FAQ Matching | |
| Bias Variance Tradeoff | |
| Backpropagation Explanation | |
| Merchant Acquisition | |
| Classification and Regression | |
| Addressing Data Quality Issues | |
| Resumable Fact Table Load | |
| Target Indices | |
| Deciding Between Solutions | |
| User Event Data Pipeline | |
| Real-Time Hashtag Partitioning | |
| String Palindromes | |
| Why Do You Want to Work With Us | |
| RAG Hallucinations |
Synthesized from candidate reports. Individual experiences may vary.
The candidate reports applying through LinkedIn and hearing from several HR contacts who said they would schedule an interview. About two weeks of silence followed before one of them booked a time. In this account, scheduling was slow, so keeping in touch with recruiters is worthwhile.
After a two-day reschedule caused by the interviewer's connectivity issue, the candidate describes a technical session. It covered a detailed ML project walkthrough, including preprocessing, hyperparameter tuning and metric choices. It also included writing F1 score code from scratch without sklearn.
The candidate reports this round came about a month later. Questions covered SVM support vectors, KNN, deep learning versus machine learning and CNN architecture. They also covered stemming versus lemmatization and RAG, including why to use an LLM instead of only retrieving from a vector database. The candidate received no offer.