From our research
What this interview looks like
Data, AI and enterprise-platform hiring in India typically runs an HR or recruiter screen, a hiring-manager round on past projects, one or two technical rounds (a live SQL or case discussion for analytics, a modelling and systems discussion for ML, a configuration and scenario walkthrough for SAP, Salesforce, ServiceNow and Dynamics), and a behavioural or fitment round, spread over one to three weeks. Many IT-services and GCC loops add a client interview before the offer. ExamPilot compresses this into one 17-minute panel: HR screen, hiring-manager round on what you actually built and the business result, a role-specific technical round led by a lead or principal, a STAR behavioural round and a closing on expectations. Rounds are pass/fail in real companies; here every round is scored on role knowledge, problem solving, communication and attitude, with specific feedback on the trade-offs you did and did not mention.
How it is weighted
Pass/fail per round in real life, with the technical deep dive usually decisive and a bar-raiser or director round common at product companies; here each round is scored against the corporate rubric (role knowledge 30, problem solving 25, communication 25, attitude 20).
Who this is for
ML engineers, applied data scientists and GenAI or LLM application engineers interviewing at product companies, AI startups, GCCs and IT-services AI practices. Loops usually include an ML fundamentals and system-design discussion and a deep dive on one model or LLM application you took to production.
The panel focuses on
- ML lifecycle in production: leakage, skew, drift and retraining
- Choosing and defending an evaluation metric tied to a business cost
- RAG and LLM application design with evaluation and guardrails
- Explaining a model decision to a non-technical stakeholder
- Knowing when a simpler method or no model is the right call
- STAR answers on a model that failed and on disagreeing with a product manager
Step 1
Pick your domain
The expert on the panel probes these areas at the level of this role. Optional here; you can choose during setup.
Step 2
Choose how you want to be interviewed
1:1 interview
One interviewer runs every round and adapts the focus as you go. Lower pressure; ideal for a first attempt.
Ms. Priyanka Raghavan
Hiring manager
Panel interview · 3 members
Most realisticEach member leads their own round and hands over to the next. They hear each other, so a weak answer will be revisited.
Ms. Neha Kulkarni
HR member
Ms. Priyanka Raghavan
Hiring manager
Mr. Arjun Nair
Technical lead
What to expect
5 rounds · 17 minutes
A private-sector hiring loop compressed into one sitting, the way most Indian companies run it: HR screen → hiring-manager round on your actual work → role/technical round led by a lead → communication & behavioural (STAR) round → closing on expectations and your questions. Works for software, product, sales, finance, HR, support and campus hiring by swapping the role bank.
- Round 1
HR screen
· 3:00Understand background, motivation for this role and company, and basic logistics; check that the resume story holds together.
Ms. Neha Kulkarni2–3 questionsup to 1 follow-up each - Round 2
Hiring-manager round
· 4:00Assess ownership, judgement and how the candidate actually works: what they shipped, closed or ran, their specific part, and how they would approach this job.
Ms. Priyanka Raghavan2–3 questionsup to 2 follow-ups each - Round 3
Role / technical round
· 5:00Test role knowledge and problem-solving in the chosen specialisation: fundamentals, one scenario to think through aloud, and trade-offs.
Mr. Arjun Nair3–4 questionsup to 2 follow-ups each - Round 4
Communication & behavioural round
· 3:00Assess collaboration, conflict handling, resilience and clarity of expression through Situation–Task–Action–Result examples.
Ms. Priyanka Raghavan2–2 questionsup to 1 follow-up each - Round 5
Expectations & closing
· 2:00Cover practical expectations and let the candidate ask questions; close professionally.
Ms. Neha Kulkarni1–2 questions
Sample questions
Data, AI & enterprise AI & ML interview questions
Questions panels commonly ask in the Data, AI & enterprise AI & ML interview, with what a strong answer covers. Practise them aloud with the AI panel.
- Production ML
Your churn model performed brilliantly offline but poorly in production. What could have gone wrong?
What a strong answer covers
Discuss data leakage from future information in features, training-serving skew, distribution drift, label definition changes and pipeline bugs. Explain how you would check each, such as comparing feature distributions. Avoid assuming the model simply needs more data or a bigger algorithm.
- Evaluation
For a loan-default model at an NBFC, would you optimise for precision or recall, and how would you choose the threshold?
What a strong answer covers
Tie the choice to business costs: missed defaulters versus rejected good customers. Use a cost matrix, plot precision-recall and choose a threshold that meets risk appetite, possibly with manual review bands. Mention fairness checks. Avoid choosing accuracy on an imbalanced dataset.
- RAG design
Design a retrieval-augmented assistant that answers employee HR-policy questions for a company with five thousand pages of documents.
What a strong answer covers
Cover chunking, embeddings and a vector store, retrieval with metadata filters, prompt design with citations, access control, evaluation sets with expected answers, guardrails for out-of-scope questions and monitoring. Mention refresh when policies change. Avoid skipping evaluation or letting the model answer without sources.
- LLM evaluation
How would you evaluate whether an LLM-based customer support bot is good enough to go live?
What a strong answer covers
Build an evaluation set of real queries, measure correctness, groundedness, tone and escalation behaviour, use human review plus automated scoring, and run a limited pilot with metrics like resolution rate and CSAT. Set clear go-live thresholds. Avoid going live on a handful of demos.
- Explainability
Explain to a non-technical operations head why your model rejected a particular delivery partner's application.
What a strong answer covers
Use plain language: the main factors that influenced the decision, using explainability tools like SHAP translated into simple reasons, and what could change the outcome. Mention a review path. Avoid jargon such as feature importances or saying the model just decided.
- Simplicity
A product manager wants a deep learning model to predict daily demand, but a simple moving average performs nearly as well. What do you recommend?
What a strong answer covers
Compare accuracy gains against cost, latency, maintenance and explainability, and recommend the simpler method if the improvement is not material. Suggest monitoring and revisiting later. Avoid using complex models for their own sake.
- Monitoring
How do you detect and respond to data drift in a model that is already in production?
What a strong answer covers
Monitor feature and prediction distributions using statistical tests, track performance with delayed labels, set alerts, investigate the cause, and retrain or roll back as needed. Mention a retraining pipeline. Avoid retraining blindly on a schedule without checks.
- Failure
Describe a model or AI feature you built that failed to deliver value. What did you learn?
What a strong answer covers
A STAR answer: the goal, approach, why it failed such as poor problem framing, data quality or lack of adoption, how you discovered it and what you changed in your approach afterwards. Avoid blaming stakeholders.
- Stakeholders
You disagreed with a product manager about launching a model that you thought was not ready. How did it go?
What a strong answer covers
A STAR story: your concern with data, the PM's priorities, the compromise such as a limited rollout with monitoring, and the outcome. Show collaboration. Avoid presenting yourself as the only one who cared about quality.
Reading is not rehearsing. Answer these out loud to an AI Data, AI & enterprise AI & ML panel that follows up like the real one.
Practise these questionsMeet the panel
Your AI interviewers
Distinct personas, voices and questioning styles, briefed on this role.
Ms. Neha Kulkarni
ChairHR member
HR business partner, IT services
Friendly, efficient, detail-checking.
Female voice · 4 languages
Ms. Priyanka Raghavan
Hiring manager
Head of Data & Analytics, consumer internet company (hiring manager for data, analytics and ML teams)
Sharp, numbers-first, allergic to dashboards nobody acts on.
Female voice · 3 languages
Mr. Arjun Nair
Technical lead
Principal Machine Learning Engineer, AI platform team (technical interviewer for ML, data science and GenAI roles)
Calm, rigorous, enjoys a good trade-off discussion more than a right answer.
Male voice · 3 languages
Languages
Answer in the language you think in
The panel asks in your chosen language. Switch mid-answer if you like.
- English
- Hinglish (Hindi + English)Hinglish
How scoring works
A report you can act on
Every round is scored on the rubric
Each interviewer scores only the criteria their round covers. Weights add up to your overall score out of 100.
Only what you actually said counts
Feedback quotes your own answers. Rounds you skip show as “Not assessed” rather than a zero.
Communication is always measured
Fluency, clarity, confidence and structure are tracked across the whole interview, in any language.
Pass mark, then a plan
You see the pass mark for this interview, your gaps, and the courses that close them fastest.
Scored on
- Role knowledge30%
Depth and accuracy of role/domain knowledge; understands the tools, concepts and trade-offs of the job; grounds claims in real work.
- Problem solving & structure25%
Breaks problems down, asks clarifying questions, reasons about trade-offs and edge cases, structures answers (context, action, result).
- Communication25%
Clear, concise, confident; listens and answers the question asked; adapts the explanation to the listener; professional register.
- Culture & attitude20%
Ownership, honesty about limits, coachability, collaboration, motivation for this role and realistic expectations.
Pass mark 65 / 100 · Corporate interview — HR, hiring manager & role round (panel)
Keep exploring