Company Interview Questions · Big Tech
Oracle & Salesforce: AI / ML Engineer
Oracle & Salesforce: AI / ML Engineer Interview Experiences (2025). Rounds, real reported questions and model answers, compiled from public candidate reports (sources at the end). Practice mode hides every answer: say your answer first, press Show answer, then rate yourself. Use 🔁 Review again to build your revision list and 🎲 Random question for a quick drill.
About this interview
Compiled from public candidate reports: Oracle Software Engineer – AI (Pune, November 2025, offer) and Oracle AI Engineer (May 2025), plus a public process guide for Salesforce ML Engineer roles. The questions are paraphrased, and the model answers are written by EasyLearnSpace.
| Oracle | Salesforce | |
|---|---|---|
| Role | Software Engineer – AI; AI Engineer | Machine Learning Engineer / AI Engineer |
| Rounds | HackerRank (AI, LLMs, aptitude, coding) → face-to-face coding → managerial. AI Engineer: online assessment (MCQs on ML, SQL, probability + 2 Python problems) → VP one-to-one | HackerRank (about 3 h, Python + ML) → HM intro → about 5 onsite rounds (design, coding, managerial) → possible project presentation |
| Reported questions | String rotation, anagram check, SQL self-join, identical binary trees; why are you switching? | RAG and its uses, LLM experience, monitoring models in production, prompt engineering approach |
| Outcome | Pune SE-AI: offer accepted | Guide-based (no single report) |
01Rounds at a glance
- Oracle: classic DSA + SQL plus AI MCQs, then managerial. Solid fundamentals win.
- Salesforce: ML system design + coding + a presentation, with a focus on enterprise AI (Agentforce/Einstein), trust and data security.
02Questions and model answers
Q1
Check whether one string is a rotation of another.
def is_rotation(a: str, b: str) -> bool:
return len(a) == len(b) and b in (a + a)
print(is_rotation("waterbottle", "erbottlewat")) # True
- Every rotation of
ais a substring ofa + a. That is O(n) with KMP, and Python'sinis efficient in practice.
Q2
Check whether two binary trees are structurally identical (same shape and values).
def same_tree(p, q):
if not p or not q: return p is q
return p.val == q.val and same_tree(p.left, q.left) and same_tree(p.right, q.right)
- O(n) time, O(h) recursion depth. For "structure only", drop the value comparison.
Q3
SQL self-join: list employees alongside their manager's name.
SELECT e.name AS employee, m.name AS manager
FROM employees e
LEFT JOIN employees m ON m.emp_id = e.manager_id;
- A self-join joins a table to itself with aliases. Use LEFT JOIN so the CEO (no manager) still appears. A follow-up asks for employees who earn more than their manager: add
WHERE e.salary > m.salary.
Q4
Salesforce: explain RAG and where it fits in a CRM.
- RAG grounds LLM answers in retrieved enterprise data, which reduces hallucination and keeps answers current.
- CRM uses:
- A service agent answers cases from knowledge articles and past resolutions.
- Sales gets account summaries from opportunities, emails and calls.
- Field-service troubleshooting.
- Trust layer: respect record-level sharing rules at retrieval, mask PII before the LLM, apply zero-data-retention with the model provider, log audits, and filter toxicity. Cite the source records.
Q5
How do you evaluate and monitor ML or LLM models in production?
- Online metrics: business KPIs (case deflection, conversion), user feedback, latency and error rates, cost per request.
- Model health: input and output drift, prediction-distribution shifts, data-quality checks, performance on delayed labels.
- LLM-specific: sampled LLM-judge scores (faithfulness, helpfulness), refusal and toxicity rates, prompt-injection attempts, retrieval hit rate.
- Process: dashboards, alerts against SLOs, shadow and canary releases, periodic re-evaluation on a golden set, rollback plans.
Q6
Oracle managerial: why do you want to switch companies?
- Stay positive and forward-looking: "I've built strong foundations in X at my current company. I'm looking for larger-scale AI problems, like the OCI GenAI services and AI features in Oracle Database (vector search, Select AI), where I can work closer to the platform."
- Avoid criticising your current employer. Connect your skills to the team's work.
- Have a clear expected salary range, based on research, ready.
03How to prepare
- Oracle: revise DSA basics (strings, trees), SQL joins and self-joins, and AI MCQs (ML, probability, LLM basics).
- Salesforce: prepare an enterprise RAG/agent design with a trust layer, plus a 15-minute project presentation.
Sources
- Taro: Oracle Software Engineer – AI, Pune (Nov 2025, offer)
- Glassdoor: Oracle AI Engineer interview reviews
- Interview Query: Salesforce ML Engineer guide
Created by Ajeetkumar