
Company interview guide
Bank of America Machine Learning Interview Guide (2026)
A conventional bank loop: recruiter and hiring-manager calls, a technical conversation, sometimes a take-home, with statistics-heavy questions for quant roles.
Bank of America's machine learning interview loop has 7 reported stages, from application review & recruiter call to final interviews, values & offer; typical length: Not well reported; the company says a decision takes 2-4 weeks after the business interview.
Roles covered: Data Scientist, ML Engineer and Applied Scientist. The question bank holds 17 questions reported from Bank of America interviews, 7 of them free to solve in your browser.
- Interview process length
- Not well reported; the company says a decision takes 2-4 weeks after the business interview
- Stages
- 7
- Reports linked to bank questions
- 17
- Bank report dates
- Jun 2016–Jun 2026
- Questions in bank
- 17 (7 free)
- Last reviewed
- October 2026
How the Bank of America interview loop works
Experienced hires apply to a specific role; a talent acquisition partner may screen you before the hiring manager reviews your profile. One 2022 data-science report ran a recruiter phone screen, a call with the hiring manager and a lead engineer, a technical call with an engineer, then a take-home modelling project before the offer. Some roles add an emailed online assessment, which can be retaken only after six months. Campus programs (summer analyst, full-time analyst and associate) are reported in compiled accounts to start with a timed behavioral and judgment assessment and a recorded video interview, then a final round of back-to-back interviews where quantitative teams add math and programming. Quant and modelling teams appear to test statistics and time series more than deep learning. Few first-hand reports exist, so treat the order as a guide rather than a fixed loop.
The Bank of America interview process, stage by stage
01Application review & recruiter call
Phone call with a recruiter or talent acquisition partner; length not reported
You apply to a named role in a business line. A talent acquisition partner contacts you if you are selected, and may or may not hold a screening call before passing top applicants to the hiring manager. Your profile can also be considered for other open roles.
What it tests
- Fit between your background and the specific team's work
- Motivation for banking and for this business line
- Logistics: location, level, timing
How to prepare
- Read the job posting closely; the company says technical requirements live in each posting, not on program pages.
- Ask which later steps apply to your role: an assessment, a take-home, how many interviews.
One 2022 data-science report starts with a recruiter phone screen. The company's hiring page and FAQ describe contact from talent acquisition, who may screen before the hiring manager review.
02Online assessment & video interview (campus and some roles)
Emailed timed assessment for some roles; campus candidates often also record video answers. Compiled campus accounts put the assessment at about 30 min
The company says some roles require an assessment, sent by email, that measures aptitudes and values; failed assessments can be retaken after six months and results are not shared. Compiled campus accounts describe it as a behavioral and judgment battery with a short numerical business case, not a math or coding test, followed by recorded answers to standard behavioral prompts. No first-hand ML or quant report in our set mentions it.
What it tests
- Workplace judgment: ownership, prioritising, raising problems early
- Reading a short business case and doing quick arithmetic on it
- Consistent self-description across personality items
- Clear, structured answers to motivation and teamwork prompts
How to prepare
- Take it somewhere quiet and in one sitting, as the company advises.
- Answer the personality items honestly and consistently; compiled accounts say near-duplicate items check consistency.
- Prepare short teamwork, conflict and why-this-bank stories for the recorded video.
The company's hiring page and FAQ say some roles require an emailed assessment of aptitudes and values. Format details come only from compiled campus accounts; no first-hand report here.
03Hiring-manager call
Call or video with the hiring manager, sometimes joined by a lead engineer
The company calls this the line-of-business interview, and it shares the hiring manager's name once you are selected. In the one first-hand data-science report the hiring manager and a lead engineer met the candidate together before any technical call, so expect questions on your past projects as well as fit.
What it tests
- Depth on past data science or modelling projects
- How you would work with the team's data and stakeholders
- Interest in the bank's products, history and mission
How to prepare
- Look up the hiring manager and the team's business line before the call.
- Prepare one project walk-through with the data, model, metric and business result.
One 2022 data-science report had a joint call with the hiring manager and a lead engineer. The company's FAQ names a line-of-business interview with the hiring manager.
04Technical interviews: coding & ML fundamentals
One or more live calls with an engineer or team member; length and tooling not reported
The first-hand data-science report had one technical call with an engineer that included a conceptual ML question (what an embedding is). Coding questions tagged to the bank are classic data-structure problems with a streaming flavour: windowed maximum, median and percentile, heavy hitters, reservoir sampling, range-sum counting and caches. Those questions were tagged to software-engineer interviews, so data-science loops may weigh them less.
What it tests
- Explaining core ML ideas such as embeddings in plain terms
- One-pass algorithms over streams with bounded memory
- Sliding-window data structures (deques, two heaps, order statistics)
- Prefix sums, binary search and Fenwick trees for range queries
How to prepare
- Be ready to define embeddings, regularisation and evaluation metrics crisply, with an example from your work.
- Practise the sliding-window trio: maximum with a deque, median with two heaps, percentile with an ordered structure.
- State time and space complexity before coding, and handle empty or tiny windows.
Free reported questions for this stage
- Sliding Window Maximum with a Monotonic DequeCoding & AlgorithmsMediumSolve
- Reservoir Sampling from an Unknown-Length StreamCoding & AlgorithmsMediumSolve
- Greedy Interval Covering of Sorted PointsCoding & AlgorithmsMediumSolve
- 2D Fenwick Tree for Rectangle Sum QueriesCoding & AlgorithmsHardSolve
One 2022 data-science report describes a technical call with an engineer. The coding problems here come from software-engineer interview tags without round details and are not counted.
05Quant & modelling rounds: statistics, regression & time series
Quant and modelling roles; format not reported first-hand, compiled campus accounts mention math and programming interviews in the final round
Questions tagged to quant-researcher interviews at the bank are derivation-heavy applied statistics: regression with serially correlated errors, autoregressive order selection, change detection on returns, Bayesian regression, model-selection criteria and ROC AUC as a ranking probability. Expect to state assumptions, derive an estimator or test, then explain how you would use it on real data.
What it tests
- Regression diagnostics and fixes when errors are autocorrelated
- AR models: Yule-Walker, partial autocorrelation and choosing the order
- Sequential change detection such as CUSUM
- Bayesian linear regression and model-selection criteria
- Interpreting classifier metrics probabilistically
How to prepare
- Rehearse deriving GLS and Cochrane-Orcutt from the AR(1) error model, not just naming them.
- Know how AIC, BIC and adjusted R-squared trade fit against complexity.
- Tie each derivation back to risk or model validation, which is what modelling teams at a bank do day to day.
Free reported questions for this stage
- GLS and Cochrane-Orcutt for AR(1) ErrorsStatistics & InferenceHardSolve
- Similarities Between Adjusted R-Squared and AICStatistics & InferenceEasySolve
No first-hand round description. Problems come from quant-researcher interview tags (2016-2026) without round details; compiled campus accounts say quant finals include math and programming.
06Take-home modelling project (some teams)
Technical take-home after the live interviews in one report; length and tooling not reported
In the one first-hand data-science report the last step before the offer was a take-home: choose useful features from a set of input columns and predict a target. Treat it as a small end-to-end modelling exercise judged on reasoning as much as on accuracy.
What it tests
- Exploratory analysis and data cleaning on tabular data
- Feature selection with a clear rationale
- Choosing a baseline and a validation scheme
- Communicating results and limitations
How to prepare
- Start with a simple baseline and show how each feature choice changes validated performance.
- Write up assumptions, leakage checks and what you would try next, not just the final score.
One 2022 data-science report ended with a take-home on feature selection and prediction; no other report confirms it, so it may be team-specific.
07Final interviews, values & offer
In-person or video conversations with business managers and peers; campus finals run as back-to-back interviews in compiled accounts
The company says its interviews cover the bank's products, services, history and mission and how your aspirations and values align with them, and that offers end with a signed contract letter. Campus finals are described in compiled accounts as several back-to-back behavioral and division-specific interviews. The one first-hand experienced-hire report went from the take-home straight to an offer.
What it tests
- Knowledge of the bank's businesses and the team's role in them
- Values: integrity, teamwork, leadership
- Motivation for the specific business line
How to prepare
- Learn the bank's lines of business and where your target team sits among them.
- Prepare examples of teamwork, integrity and leadership, which the company names directly.
- Expect 2-4 weeks between the business interview and a decision.
The company's hiring page and FAQ describe interviews with business managers and peers, a values focus and a 2-4 week decision; campus final-round details come only from compiled accounts.
Questions and themes reported most often
- Sliding-window and streaming statistics: maximum, median, percentile, heavy hitters, reservoir sampling (3 reports, technical interviews: coding & ml fundamentals)
- Range and order queries on arrays: subarray sums in a range, k-th pair sum, 2D prefix sums, LIS (3 reports, technical interviews: coding & ml fundamentals)
- Regression and time series with correlated errors: GLS, AR order selection, CUSUM (2 reports, quant & modelling rounds: statistics, regression & time series)
- Bayesian regression and model-selection criteria (2 reports, quant & modelling rounds: statistics, regression & time series)
- Explain what an embedding is (1 reports, technical interviews: coding & ml fundamentals)
- Take-home: select features from tabular columns and predict a target (1 reports, take-home modelling project (some teams))
- ROC AUC as the probability a positive outranks a negative (1 reports, quant & modelling rounds: statistics, regression & time series)
- Design an LFU cache (1 reports, technical interviews: coding & ml fundamentals)
What Bank of America asks about
- Coding & data structures30%
- Statistics, regression & time series30%
- ML & data science fundamentals15%
- Modelling take-home10%
- Behavioral & values15%
Bank of America interview prep checklist
- Asked the recruiter which assessment, interviews and take-home apply to the role
- Can explain embeddings, regularisation and core metrics with a project example
- Solved sliding-window maximum, median and percentile with correct complexity
- Derived GLS for AR(1) errors and Yule-Walker for AR order selection
- Built a small feature-selection and prediction notebook with a clear write-up
- Prepared why-this-bank and values stories tied to the target business line
- Reviewed the bank's lines of business and where the team fits
Tips
- Ask your recruiter early which steps your team uses; the bank's process is run per business line and varies.
- For data-science roles, prepare a tabular modelling workflow you can run quickly: baseline, feature selection, validation, write-up.
- For quant and modelling roles, prioritise applied statistics and time series over deep learning.
- Practise streaming and sliding-window data structures; they dominate the coding questions tagged to the bank.
- Know the bank's products, history and mission; the company says interviews test this directly.
- Have values stories ready on integrity, teamwork and leadership.
- If an emailed assessment arrives, do it in one quiet sitting; a failed attempt locks you out for six months.
- Treat this outline as provisional: it rests on very few first-hand reports.
See the rest of the Bank of America loop
The later stages, the questions reported for them and how to prepare are in Pro.
Unlock with ProPractice the Bank of America loop
Work the stages in order: every reported question for this company, grouped by round, with solutions in the app.
Bank of America interview FAQ
How many interview rounds does Bank of America have for ML roles?
Candidates report 7 stages: Application review & recruiter call, Online assessment & video interview (campus and some roles), Hiring-manager call, Technical interviews: coding & ML fundamentals, Quant & modelling rounds: statistics, regression & time series, Take-home modelling project (some teams) and Final interviews, values & offer. Experienced hires apply to a specific role; a talent acquisition partner may screen you before the hiring manager reviews your profile. One 2022 data-science report ran a recruiter phone screen, a call with the hiring manager and a lead engineer, a technical call with an engineer, then a take-home modelling project before the offer. Some roles add an emailed online assessment, which can be retaken only after six months. Campus programs (summer analyst, full-time analyst and associate) are reported in compiled accounts to start with a timed behavioral and judgment assessment and a recorded video interview, then a final round of back-to-back interviews where quantitative teams add math and programming. Quant and modelling teams appear to test statistics and time series more than deep learning. Few first-hand reports exist, so treat the order as a guide rather than a fixed loop.
How long does the Bank of America ML interview process take?
Reported timelines run Not well reported; the company says a decision takes 2-4 weeks after the business interview.
What does Bank of America ask in ML interviews?
By reported question mix: Coding & data structures 30%; Statistics, regression & time series 30%; ML & data science fundamentals 15%; Modelling take-home 10%; Behavioral & values 15%. The most repeated themes: Sliding-window and streaming statistics: maximum, median, percentile, heavy hitters, reservoir sampling; Range and order queries on arrays: subarray sums in a range, k-th pair sum, 2D prefix sums, LIS; Regression and time series with correlated errors: GLS, AR order selection, CUSUM.
How should I prepare for the Bank of America ML interview?
Ask your recruiter early which steps your team uses; the bank's process is run per business line and varies. For data-science roles, prepare a tabular modelling workflow you can run quickly: baseline, feature selection, validation, write-up. For quant and modelling roles, prioritise applied statistics and time series over deep learning.
Are there free Bank of America interview questions?
Yes. LeetML has 17 questions reported from Bank of America interviews, and 7 of them are free to solve in the browser; they are listed by stage above.
Related interview guides
- JPMorgan ChaseA recorded first-round assessment, then team interviews that test ML basics, your own projects and, for quant tracks, probability and statistics.
- CitiA thinly documented loop: an application and skills check, a background interview, then a final stage of technical interviews, with quant roles leaning on probability and markets maths.
- Capital OneA timed four-question coding assessment, then a four-round Power Day of ML technical, a business case, behavioral and, on recent ML engineer loops, ML system design.