RepostedJP Morgan Chase logo

JP Morgan Chase · OA → Phone screen → Onsite

In process

JP Morgan Chase Data and AI Intern Superday: 11 Experiences

Data science, AI and NLP internships · InternshipReported Oct 17, 2026
Shared byjanebeingjane

JPMorgan Chase data and AI intern reports, 2023 to 2026: Superdays of three 45-minute rounds on ML concepts, projects and motivation, with little coding.

At a glance

  • Process: HireVue and an online skills test or coding tasks, sometimes a screen, then a Superday of two or three interviews of about 45 minutes (02, 03, 04, 05).
  • Asked recently: ML concepts, metrics or projects in 4 of 5 current reports; resume and project talk in 3; behavioral and motivation in 3. Named coding tasks came only from one HireVue (02).
  • Difficulty: the 2026 screen had no coding and felt accessible (01); the 2025 Superday stayed conceptual with no coding (02); a Quant AI candidate said both ML rounds went badly (05).
  • Practise first: Decode Ways and Population Standard Deviation, both from the same HireVue (02).

Coverage: JPMorgan Chase AI and Data Science, Data Science Analyst, Quant AI and a VR/ML summer-associate internship, from screen to Superday. Five current reports from November 2023 to October 2026, plus six older reports (2021–2023) under History, including AI Research, NLP and an Indian campus route. Wilmington, New York and India where stated.

What came up

The current reports lean on talk more than code: ML concepts, the candidate's own projects and why they want the role. Only one report named coding tasks, and those came before the Superday.

Topic Reports Last seen
ML concepts, metrics or ML projects 4 of 5 Oct 2026 (report date)
Resume, project or research discussion 3 of 5 Oct 2026 (report date)
Behavioral and motivation 3 of 5 May 2025 (report date)
Algorithms (concepts or coding) 2 of 5 May 2025 (report date)
SQL and Python questions 1 of 5 Oct 2026 (report date)
Case discussion 1 of 5 May 2025 (report date)
Finance and markets 1 of 5 Jan 2025
Data analysis skills test 1 of 5 Jan 2025

A candidate can appear in more than one row.

Questions we found

Algorithms / Data Structures

Question Reports Last seen
Decode Ways 1 May 2025 (report date)
Standard deviation of an array 1 May 2025 (report date)

Behavioral (BQ)

Question Reports Last seen
Why this role? 1 Jan 2025
Walk through points on your resume 1 Jan 2025

Finance and markets

Question Reports Last seen
What markets do you follow? 1 Jan 2025
What current factors affect financial services? 1 Jan 2025

Research presentation

Question Reports Last seen
Present VR game algorithms combined with ML, then take questions 1 Dec 2023 (report date)

5 detailed candidate timelines, loops, and screens

01: A data and AI screen with no coding

Report date: October 17, 2026 · Role: Data and AI intern · Stage: Screen

  1. The screen opened with the candidate explaining their ML projects.
  2. SQL, Python and ML technical questions followed. There was no coding.
  3. They found the technical questions accessible but were less sure they had explained their projects clearly. No single question was written down.

02: Three 45-minute Zoom sessions and no code

Recruiting cycle: Summer 2025 · Report date: May 29, 2025 · Role: AI and Data Science summer associate intern · Result: No offer

HireVue

  1. The candidate first recalled three or four tasks, then described two simple technical tasks and two medium data-structure tasks.
  2. Two were named: the standard deviation of an array (they did not say population or sample) and Decode Ways.

Superday

  1. Three 45-minute Zoom sessions: behavioral, technical and a case discussion.
  2. They had been told to prepare SQL and data structures, but the technical session stayed on algorithm concepts, metrics and equations, with no coding. The case goal and the technical questions were not shared.
  3. An interviewer mismatch led to a reassignment along the way. The whole process ran more than three months.

03: Markets and motivation in Wilmington

Interview month: January 2025 · Role: Data Science Analyst intern · Location: Wilmington, US · Result: No offer

  1. A short HireVue, then a skills test with a choice of language. It covered extracting, organizing and analyzing data; the exact operations were not shared.
  2. Superday: why this role, which markets they follow, which current factors affect financial services, and points on their resume.
  3. No session lengths or coding problem were given. The application took two to three months.

04: A 45-minute VR and ML presentation

Report date: December 5, 2023, with interview details added January 6, 2024 · Role: Summer associate, VR direction with ML · Result: Offer

  1. Superday, 90 minutes with three interviewers: the candidate presented VR game algorithms combined with machine learning for 45 minutes.
  2. The next 45 minutes were questions on the project and possible directions for the internship, plus a good deal of behavioral discussion. The behavioral prompts were not listed.
  3. They guessed the three interviewers would be their future team; that was not confirmed. The start date was expected in June 2024. Whether they accepted was not stated.

05: Quant AI: one algorithm round, two ML rounds

Interview date: November 30, 2023 · Role: Quant AI intern

  1. Three Superday rounds: an algorithm round first, then two machine-learning rounds.
  2. None of the Green Book probability questions they had prepared for came up.
  3. By their own account both ML rounds went badly, and they expected a rejection. No decision, algorithm prompt or ML question was shared.

History (before November 2023)

H01: AI/DS: KNN, a tree path sum and a project down to running costs

Interview date: October 10, 2023 · Report date: October 11, 2023 · Role: AI/DS summer 2024 intern

  1. An engineering PhD student with four earlier ML internships. They applied in July, passed a technical screen and got the Superday invite on October 2. Three 40-minute rounds.
  2. Round 1: a commercial-bank interviewer spent about 20 minutes on two past internships, then asked the difference between KNN and K-means. After the answer the interviewer said that part was passed, and the rest was a chat about past work and large language models.
  3. Round 2: the interviewer from the earlier screen asked them to explain the law of large numbers, then for the largest path sum in a binary tree, from any node to any node. Both took about ten minutes. The rest covered Transformers and JPMorgan's work, and the round ended about ten minutes early. No code or approach was shared.
  4. Round 3: a retail-bank interviewer asked about loss functions, classification versus clustering, overfitting, parameter tuning, cross-validation for judging a model, true and false positives, precision and recall, and how to balance them with imbalanced data. A conditional-probability question followed that the candidate solved with Bayes' formula; the numbers were not shared.
  5. The last 25 minutes or so were on a past internship project. It overlapped with work a partner team was planning, so the follow-ups went deep, down to the running cost after launch. In a later reply the candidate also listed ML theory, tuning, system design and offline, online and production testing as project topics, without tying each to a round.
  6. HR called on October 24. Two hiring-manager calls on October 26 and 27 were team introductions, not interviews. The team was confirmed October 30, with a formal offer expected within a week. They held another offer and had not accepted at the time.

H02: ML basics, behavioral and a deep project interview

Interview date: January 24, 2023 · Report date: February 22, 2023 · Role: AI and Data Science summer associate, summer 2023 · Location: New York team · Result: Formal offer

  1. Applied August 17, 2022, got the OA the next day and finished it August 22.
  2. The Superday was three back-to-back video interviews: ML basics, behavioral, and project experience. The project interview went deep with many follow-ups. No prompts were listed.
  3. Verbal offer January 30; talks with three teams January 31; formal offer February 6. They chose a team in the New York office. Acceptance was not stated.

H03: AI Research: two research interviews and a false-negative case

Interview date: November 16, 2022 · Report date: December 2, 2022 · Role: Summer AI Research intern · Result: Offer accepted

  1. Invited November 7. On November 16, two 45-minute interviews centered on their research.
  2. One interviewer added a case study about false negatives. The scenario, data and decision asked for were not shared.
  3. Verbal offer November 22, written offer November 25, accepted before the December 1 deadline.

H04: A campus mentorship route to a data science internship

Publication edition: 2022 · Role: Data scientist intern · Location: IIT Kharagpur campus recruitment, India · Result: Internship received

  1. Mentorship selection: multiple choice on entrance-exam-level math, probability and statistics, data structures, OOP and Python, then two programming-style cases. One was about options, without the operation asked for; the other was about cellular automata, without its rules, grid, starting state or required output.
  2. Internship interviews, around mid-July: four rounds. The first three covered probability and statistics, puzzles, ML and computer vision; the fourth covered data structures and Python. The exact year and day of these interviews are not known.

H05: An NLP interview that went past NLP

Interview month: January 2022 · Role: NLP / ML intern · Result: Rejected in February

  1. A recruiter called about the internship. The interview was set for around January 4, after about two weeks of preparation.
  2. The candidate expected NLP. The interview also covered coding, statistics, ML frameworks and the resume. No exercise was named.

H06: Three easy coding questions for an AI summer associate role

Report month: August 2021 · Role: AI summer associate intern, summer 2022 program · Stage: Assessment

  1. The invite came the week before. The assessment had three basic coding questions and one behavioral question.
  2. They later called the coding LeetCode easy. No decision had arrived, and no prompt was shared.

Patterns across the reports

  1. Superdays come in threes. Three current reports had three interviewers or rounds (02, 04, 05), and so did two History reports (H01, H02). Only the AI Research candidate had two (H03).
  2. Projects take up real time. Candidates were asked to explain or present their own work in 01, 03 and 04, and the project rounds in H01 and H02 drew long follow-ups.

Common questions

Does the JPMorgan Data and AI intern Superday include coding?

Not in the two recent reports that said: the 2025 Superday stayed conceptual (02) and the 2026 screen had no coding (01). Coding showed up earlier, in HireVue (02).

How many interviews are in the JPMorgan AI and Data Science Superday?

Usually three, of 40 to 45 minutes each (02, 05, H01, H02). The VR/ML candidate had one 90-minute session with three interviewers (04).

What ML questions does JPMorgan ask data and AI interns?

The fullest list is from 2023: KNN versus K-means, loss functions, overfitting, cross-validation and precision versus recall with imbalanced data (H01). Recent reports mention ML questions without listing them (01, 05).

How long does the JPMorgan data science intern process take?

Two to three months for the Data Science Analyst candidate (03) and over three months for the 2025 AI and Data Science candidate (02).

Keep practicing

Sources: Anywhere useful interview info can be found.

0 total reactions. Your reactions: none.
0 comments

Comments

0

No comments yet — be the first to share what you know.