[New Grad] Hyundai Motor — CarCloud Role — Coding Test + 1st Interview (Rejected) Recap (2023 H1)

📅 Hyundai Motor 2023 H1 new-grad recruitment — CarCloud role recap. Result: rejected at the 1st interview on 2023-05-12. I self-graded 4/6 well-answered, 1 wrong, 1 unknown. The gap between self-grading and outcome is what makes this one stick.

Timeline

Aptitude + AI interview (2023-04-04)

Slot: 07:00–22:00, candidates choose their time.

A. Aptitude (1h)

Standard personality assessment. Not hard.

B. AI interview — ViewInter platform

12 questions. Per question: 30–60s prep + 60–90s response.

• Self-intro + motivation
• Role-relevant experience
• Questions about Hyundai vehicles
• 3 thematic combo sets (3 questions each):
• Set 1: teamwork toward a shared goal
• Set 2: creative ideation experience
• Set 3: resolving team conflict / opinion clashes

Combo sets drill into one theme, so prep your stories 3-deep for each.

Result: passed both.

Coding test (2023-04-18, 10:00–12:00)

Platform: Softeer (softeer.ai) — Hyundai's own platform. 3 problems, 2 hours.

Q1 — Coordinate simulation

n×n grid, (1,1) = top-left, (n,n) = bottom-right. Numbers filled by a rule. Return number at (r,c).

Not hard per se, but the 2-problems-in-2-hours pressure masked the pattern.

Time: ~50 min.

Q2 — Combination/permutation (mathematical)

N people, M sports (N ≤ 20). Split participants in half so the skill-gap across events is minimized.

Key optimization: - N=20 → permutation = O(20!) = TLE. - Combination = O(C(20,10)) = 184,756 — fine.

Solution: pick half by combination → sum/diff across events → minimum.

Java has no built-in combinations. Implement manually. In a non-Python environment (no itertools), pre-memorize the combo generation pattern.

Time: ~1h.

Q3

Out of time, didn't attempt.

Result: passed (Q1+Q2 was enough).

1st interview — Pre-assignment + Technical

Total 40 min = 30-min pre-assignment + 10-min technical.

A. Pre-assignment (30 min, video)

• Receive problem 40 min before interview.
• Enter online waiting room 10 min before.
• 30 min to read the problem and draft a solution in an online editor.
• Present and defend to interviewers.

Topic was CarCloud's actual domain (large-scale traffic, data processing, infra). Pre-reviewing system design (LB, cache, DB sharding, async) paid off.

Drew on my then-current API Gateway project to ground the answer.

B. Technical interview (1:3)

3 interviewers : me. Follow-ups on the pre-assignment.

Pre-assignment feedback

"Logical structure was clear, problem comprehension was solid."

— Tone was good here.

Q1. API Gateway — Auth

Which auth methods have you used?

Answer: Basic Auth · mTLS from real experience.

Q2. Encryption algorithms ❌

Which encryption algorithms do you know?

Answer: RSA512. — Looking back, RSA512 itself is a weak key size (RSA2048+ is current recommendation). Wrong answer.

Q3. High-traffic data processing

• 3.1 Async processing approach?
• 3.2 Non-blocking implementation details?

Answer: Spring Cloud + WebFlux + embedded Netty — event-loop async.

Q4. OpenAPI / Swagger version compatibility

Swagger 2.x ↔ 3.x ↔ SpringBoot compat?

Answer: shared real version-compat troubles I'd hit. Fine.

Q5. High-traffic infra + cross-team collaboration

• Load balancing · distributed API Gateway.
• Coordination across network, DBA, dev.

Answer: fine.

Q6. Other languages — Python threading ❓

You said you also use Python. Java threading vs. Python threading differences?

Answer: only used Python for RL in grad school. Admitted I didn't know GIL details.

Couldn't fully articulate why Python multithreading can't use multiple cores (the GIL) — that one cost me.

C. Closing

In the open Q&A I mentioned visiting Hyundai's 2023 World IT Show booth + discussing infotainment trends + framing my motivation as "wanting to ride Hyundai's current tech wave."

Felt like a good close.

Final result

Rejected at 1st on 2023-05-12 11:25.

6 main questions: 4 well-answered (self-grade), 1 wrong (encryption), 1 unknown (Python GIL). I'd given myself passing odds. Reality: no. The self-grade vs. result gap was the lesson.

Takeaways — for cloud/MSA roles

• Pre-assignment: pre-load system design (LB, cache, sharding, async). Weave in your actual project experience.
• Encryption/security: know key-size + algorithm recommendations precisely. RSA2048+ · AES256 · TLS1.3.
• Language comparisons: know runtime internals (Python GIL, Java GC).
• Spring ecosystem: WebFlux, Netty, MSA distribution basics.
• OpenAPI / Swagger: 2.x vs 3.x, springfox vs springdoc-openapi.
• Confidence ≠ outcome: feeling good doesn't mean you passed. Don't fixate on one company.

Good luck if you're on the same path.

---

📦 Migrated from my own Korean blog (my own writing). Original: taehyuklee.tistory.com/6