Interview prep

Open access Question bank, mock practice (type or speak), stories & guides

55 questions · 19 behavioral · 24 technical · 9 guides

Suggested study order

1) Core section: STAR behavioral — prep 2 stories per theme 2) Your field section: practice by technical / product / design groups 3) Filter + mock practice; review history in Practice log 4) Before applying, read HK internship + OA guides below

All Technical (24)

Core essentials (all internships)

1 questions

Technical · 1

Whiteboard: reverse a linked list. What do you say while coding? Technical Coding basics Medium

Reference answer
Clarify constraints, brute force idea, optimal two-pointer/iterative approach, complexity, test with small example.

Tips
Communication beats perfect syntax.

Artificial Intelligence

6 questions

Technical · 6

Explain train / validation / test splits. Why not tune on the test set? Technical Artificial Intelligence Field-specific Easy

Reference answer
Train fits params; validation picks hyperparameters; test estimates generalization once. Tuning on test leaks information and overestimates performance.

Tips
Mention cross-validation for small data.

What is prompt engineering? Give one good and one bad prompt example. Technical Artificial Intelligence Field-specific Easy

Reference answer
Good: role, constraints, format, examples. Bad: vague “help me with homework” with no context.

Tips
Link to study-bot project.

Explain overfitting and how to reduce it. Technical Artificial Intelligence Coding basics Medium

Reference answer
Overfitting = model memorizes training noise. Reduce via regularization, more data, simpler models, cross-validation.

Compare precision and recall. When would you optimize each for a campus chatbot? Technical Artificial Intelligence Field-specific Medium

Reference answer
Precision = of predicted positives, how many correct; recall = of real positives, how many found. For unsafe answers, prioritize precision; for FAQ coverage, raise recall then filter.

Tips
Relate to false positive vs false negative costs.

How can ML models be biased? One mitigation in deployment. Technical Artificial Intelligence Field-specific Medium

Reference answer
Biased training data or labels; monitor slice metrics, human review on sensitive cases, diverse eval sets.

Tips
Ethics matter for HK finance/HR AI.

How would you detect data leakage in an internship ML task? Technical Artificial Intelligence Field-specific Hard

Reference answer
Look for future features, target-derived columns, improper joins, and suspiciously high metrics; validate with time-based splits.

Tips
Give one concrete example (e.g. using post-click labels).

Multimedia

2 questions

Technical · 2

What is the difference between frame rate and resolution for a short-form video internship? Technical Multimedia Field-specific Easy

Reference answer
Resolution = pixel detail; frame rate = temporal smoothness. Choose based on platform + motion; higher is not always better for file size/upload.

Tips
Mention export presets.

RAW vs compressed workflow — when does each make sense for internship deliverables? Technical Multimedia Field-specific Medium

Reference answer
RAW for color-critical re-edit; compressed proxies for fast iteration; deliver per client codec spec.

Tips
Mention storage/time trade-off.

Computer Science

13 questions

Technical · 13

What is the time complexity of binary search? Technical Computer Science Coding basics Easy

Reference answer
O(log n) on a sorted array.

What is the difference between an array and a hash map? When choose each? Technical Computer Science Field-specific Easy

Reference answer
Arrays give O(1) index access and locality; hash maps give average O(1) key lookup with extra memory and worse locality. Use arrays for ordered/indexed data; maps for sparse key lookups.

Tips
Mention collision handling briefly.

Explain the difference between == and === in JavaScript. Technical Computer Science Coding basics Easy

Reference answer
== coerces types; === strict equality without coercion. Prefer === to avoid subtle bugs.

Tips
Give one coercion example (e.g. 0 == false).

Explain REST: idempotent methods and why DELETE/PUT should be idempotent. Technical Computer Science Field-specific Medium

Reference answer
Idempotent = same request repeated leaves same server state. Clients retry safely on network failures; PUT replaces resource, DELETE removes once.

Tips
Contrast with POST creating duplicates.

How do you prevent SQL injection in a Node/Express API? Technical Computer Science Field-specific Medium

Reference answer
Never concatenate user input into SQL; use parameterized queries / prepared statements; validate inputs; least-privilege DB user.

Tips
Reference the CareerAck security lab idea.

Walk through debugging a slow API endpoint. Technical Computer Science Field-specific Medium

Reference answer
Reproduce, measure (logs/timing), check N+1 queries/indexes, payload size, external calls, then fix with evidence before/after.

Tips
Mention one tool you know (DevTools, EXPLAIN).

What happens in the browser when you type a URL and hit Enter? (high level) Technical Computer Science Field-specific Medium

Reference answer
DNS → TCP/TLS → HTTP request → server response → parse HTML/CSS/JS → render; mention cache/CDN briefly.

Tips
Stay structured; depth optional.

How would you design a simple job-application tracker API? Technical Computer Science Field-specific Medium

Reference answer
Resources: apps with company/role/status; endpoints GET/POST/PATCH/DELETE; validate status enum; persist; auth later.

Tips
Maps to the Node job-tracker lab.

What is the difference between git merge and git rebase? Technical Computer Science Coding basics Medium

Reference answer
Merge preserves branch history with a merge commit; rebase replays commits for linear history. Rebase not on shared main without team agreement.

Tips
Tie to the git-deploy lab.

Why write unit tests? What would you test in a status helper function? Technical Computer Science Coding basics Medium

Reference answer
Tests guard regressions and document behavior. Test happy path, invalid input, edge enums, and error messages.

Tips
Reference js-testing-lab.

Explain XSS and CSRF at a high level. One mitigation each. Technical Computer Science Security Medium

Reference answer
XSS = inject script in output → escape/sanitize/CSP. CSRF = forged request → tokens, SameSite cookies.

Tips
Connect to security lab.

Explain indexing in SQL. When does an index hurt performance? Technical Computer Science Coding basics Medium

Reference answer
Indexes speed selective lookups/joins; hurt on heavy write tables, wrong columns, or low-selectivity columns.

Tips
Mention EXPLAIN.

When would you use a queue or cache in a backend system? Technical Computer Science System design Hard

Reference answer
Queue for async work and peak smoothing; cache for read-heavy hot data with TTL and invalidation strategy.

Tips
Mention trade-off: staleness vs load.

Business & Analytics

1 questions

Technical · 1

Write the idea of a SQL query: count internship applications by status for each company. Technical Business & Analytics Field-specific Easy

Reference answer
FROM applications JOIN companies … GROUP BY company, status with COUNT(*); mention LEFT JOIN if companies with zero apps matter.

Tips
This mirrors the SQL lab skill.

Design & UX

1 questions

Technical · 1

Explain the difference between UI and UX with a campus app example. Technical Design & UX Design craft Easy

Reference answer
UI = visual layer; UX = end-to-end experience achieving a goal. Good UI can still have bad UX if flows fail.

Tips
Use course registration or job apply flow.

Guides Open to read

Hong Kong

HK internship interview loop

Typical loop: online application → OA / take-home (optional) → HR screen → technical or case → manager chat → offer. Prepare: 90s self-intro (EN + 粵語/普通話), 3 project stories, and questions for them. Bring: portfolio link, GitHub, and one artifact you can explain deeply.

Internship prep

60-second self-intro template (internship)

1) Name + programme + year 2) One strength with proof 3) One project relevant to the role 4) Why this team Keep under 60–75 seconds; smile and pause.

Interview guides

Interview question bank roadmap

Suggested order for interns: 1) Behavioral STAR set (general section) — 2 stories per theme 2) Your field section — technical/product/design by role tabs 3) Mock practice with filters → profile practice history 4) Read HK loop + OA guides before employer deadlines

Link interviews to your CareerAck labs

Map labs to stories: H5 portfolio = shipping UI; Kanban = state/CRUD; Node API + SQL = backend; Security lab = injection/password hashing; Deploy = ownership. Practice explaining trade-offs aloud.

Behavioral STAR cheatsheet for HK fresh grads

S/T: 1–2 sentences context. A: your actions (I, not we). R: metric or learning. Prep stories: teamwork conflict, failure, leadership without title, deadline, ethics.

OA & take-home survival tips

Read constraints twice. Start with a correct brute force if needed, then optimize. Comment assumptions. For take-homes: README, sample input, and time spent note. Never plagiarize.

STAR Method for Behavioral Interviews

Structure answers as Situation, Task, Action, Result. Keep results measurable when possible.

Technical Interview Checklist

Review DSA basics, prepare 2–3 project stories, practice aloud, and ask clarifying questions.

Technical screens

Technical phone screen — what to expect

Often 30–45 min: intro, 1–2 coding or conceptual questions, your project deep-dive, Q&A. Say your thinking aloud; ask constraints; start simple then optimize.

Experience shares

Consumer Internet — Product Design Intern

Design & UX

Portfolio walkthrough first. Then redesign critique on a messy settings page. They cared about trade-offs and accessibility, not fancy visuals only.

Local Bank Tech — Technology Summer Intern

Computer Science

HR asked STAR + why banking tech. Technical round: REST basics, one SQL group-by, and explain a project README. Tip: prepare a 60s Cantonese self-intro.

Harbor Tech — AI Intern

Artificial Intelligence

OA focused on arrays + one ML conceptual question. Onsite was system-design lite for a recommender.