American Express AI Engineer Campus Guide: DSA, Pandas, RAG and Even Clojure
Prepare for American Express AI engineer interviews with reported assessment patterns, interview questions, and preparation advice.
On this page
Last reviewed November 16, 2026.
For students applying as AI engineers.
Built by the Codinzhub team from American Express (Amex) questions and experiences that IIT students shared on Codinzhub between October 26 and November 1, 2026. Each one is one student's memory, so use this as a map, not a promise.
The 30-second version. Amex's AI engineering OA on October 31, 2026 was practical. At IIT Bombay it was 3 coding tasks: 1 DSA, 1 pandas or ML, 1 RAG (build a retrieval chatbot). At IIT Kharagpur it included a CNN image classifier, a LangGraph and LangChain chatbot and a Clojure program. Expect to write working ML and LLM code, not only algorithms.
How this guide was built. The OA comes from questions IIT students posted on Codinzhub. Where no student has posted the interview rounds yet, the Codinzhub team researched Amex's earlier campus drives, and those parts are marked.
Other Amex roles: Engineer trainee (SDE) · Technology intern · Engineer II · Data engineer · Data science and analytics · DS and analytics intern · Data analyst · AI engineer
Round-by-round breakdown
Stage 1: Online assessment
At a glance
- Length: timed; plan for about two hours
- Questions per round: 3 tasks
- What was asked: a sliding window DSA problem, logistic regression in pandas, a RAG assistant, a CNN, a LangGraph workflow, a Clojure function
- Most common: one LLM application task in every paper
IIT Bombay
| # | Question (our summary) | How to approach it | Practice |
|---|---|---|---|
| 1 | Shortest stretch of an array that contains every integer from L to R | Sliding window with counts | Shortest Fragment Containing L to R |
| 2 | Bank marketing data: predict who subscribes to a term deposit with logistic regression | Clean, encode categories, split, fit, report metrics | Predicting Term Deposit Subscriptions |
| 3 | Build a conversational RAG assistant over internal runbooks | Chunk, embed, retrieve top matches, keep chat history, answer with sources | Conversational RAG Assistant |
IIT Kharagpur
| # | Question (our summary) | How to approach it | Practice |
|---|---|---|---|
| 1 | Classify images of carrots and corn with a CNN in TensorFlow, Keras or PyTorch | Small conv net, augmentation, train and validate | CNN Carrots vs Corn |
| 2 | Extend a basic chatbot into a workflow with LangGraph and LangChain | Define nodes and edges for each step; route by state | Workflow Chatbot with LangGraph |
| 3 | In Clojure, return the second-largest distinct number, or −1 | distinct, sort descending, take the second | Clojure Programming |
More questions: Browse American Express questions.
Stage 2: Interviews
What to expect. From our research: Amex campus interviews for ML roles ask about machine learning and deep learning basics (hyperparameters, loss functions), Python and SQL, and your projects in depth, over about two rounds.
The Codinzhub Next-Round Playbook
- Build one small RAG app and one LangGraph flow before the test; setup time is the risk.
- Revise a minimal CNN training loop in your preferred framework.
- Learn enough Clojure to write simple list functions; it surprised students.
About this guide
This guide is researched and written by the Codinzhub team from what IIT students shared on Codinzhub, plus research into Amex's campus process where student reports are missing. Questions are our own short summaries; the practice links open the full question on Codinzhub. Sat this OA or the interviews? Share your experience on Codinzhub and we will update the guide.
Keep track of what you’ve learned.