[Career] Samsung Electronics DX G-CS Experienced Hire Interview (AI Vision & Data Optimization)

Hi. I want to write up a Samsung Electronics experienced-hire process I went through fairly recently — from late 2025 into early 2026.

The Role I Applied For — G-CS AI Vision & Data Optimization

The posting was open for applications from 2025.11.26 to 2025.12.08, an experienced-hire req, and the position I applied to was [G-CS] AI Vision & Data Optimization.

Application submitted.

A summary of the posting:

Responsibilities

• Build and operate an AI-powered test platform
• Auto-generate/run/judge test cases (TC) for product features using AI
• Integrate LLM/VLM AI models and build APIs
• Set up cloud (AWS·Azure·GCP) infrastructure
• Develop AI Vision and data-driven learning/prediction models
• Defect prediction and anomaly detection from manufacturing/quality data
• Image/vision-based Defect Detection, Classification, Segmentation
• ML/DL modeling from spec·log·sensor data, automation pipelines, MLOps deployment
• Large-scale data preprocessing, labeling strategy, and quality management

Requirements — 4+ years of relevant experience in AI Vision quality-management solutions, AI-based data optimization, etc.

Preferred — CS/SW/EE degree; Python·Java; AI frameworks and vision models (CNN·YOLO·SegFormer); SQL·NoSQL; hands-on AI/ML projects; OpenCV/vision processing; model testing/validation on manufacturing·quality data; big-data processing and test automation; cloud experience.

Location — Suwon, Gyeonggi · full-time. The S/W Verification Group, within the Global CS Center, builds S/W automation platforms and AI-based verification solutions and rolls them out company-wide.

Why I Thought It Was Worth a Shot

I'd done both Java and Python development, had a lot of platform-building experience, and had worked with SQL and NoSQL. On top of that, I had AI/ML reinforcement-learning research (Physical Review Fluids paper) and a prior Reinforced Security Agent I'd built at the PoC level — so I felt I had a real shot. I hadn't worked directly with vision models like YOLO, but I'd implemented my research from scratch in TensorFlow and coded some statistical models by hand, so I figured my AI/ML research experience would come across as a genuine strength.

Analyzing the role, I judged they'd hire for one of two tracks — ① building/operating the AI-powered test platform, or ② developing AI Vision/data prediction models — and I focused my cover letter on track ① (the test platform). For track ②, I framed it as "I have the experience, even if the fit isn't perfect."

Then I researched the G-CS (Global Customer Satisfaction) Center. It turned out to be closer to a QA governance org that oversees the test cases for products coming up from every business unit under the DX division — Visual Display (VD), Digital Appliances (DA), Mobile (MX), and so on (Samsung DX Recruit — team page). So I went in with at least this much of a read: "They'd want to build a test-case automation platform."

Process Timeline

Document Screening

I set my direction and wrote the application around it, and the result was a pass.

Document screening: passed.

This is where it really began. On the day I got the news, the recruiting team emailed me the full schedule.

Recruiting team's schedule email.

The overall process was announced as:

Phone interview → Technical interview (expertise / dept-head) → HR interview → Reference check & additional document verification → Offer negotiation & health check

Phone Interview

They tell you in advance which number will call, and you get scheduled with a "we'll call you around this time on this day" message.

The phone interview was announced as 30 minutes of resume-based Q&A after a team/position intro. In practice, they asked not only about the projects I'd done but also the specific problems they're currently facing. Fortunately, the questions overlapped with problems I'd actually run into in past projects. To name just the topic, it was about how files are stored in a file system (I can't go into specifics).

After about 30 minutes, the result came in about a week — interview on Dec 22, result on Dec 30.

Phone interview: passed.

Job Interview — CBT · Work Sampling · Dept-Head Interview

The next stage was the job interview + CBT (job-fit test), and passing that leads to the personality interview. The job interview was also online, run through Samsung's official video-meeting tool, Knox Meeting. They warned that noise or other people visible in frame counts as cheating, so I had to take it somewhere quiet.

The schedule came by email, and I started at 7 a.m., going CBT → job interview for about 3 hours total.

1) CBT (Job-Fit Test)

2) Work Sampling

You receive a job-related problem by email and write up an answer within a time limit (50 min), then submit it back to the recruiting team by email. They bring an actual, concrete problem they have — a Samsung-internal problem, so I can't share details. They don't demand a "correct" answer, but you do need a reasonable one. (Also run under Knox supervision.)

Honestly, this problem was quite far from the work I'd been doing (systems architecture, backend server development). I've since moved to an AI Innovation Lab within my company where I could handle this kind of problem reasonably well, but at the time it wasn't easy. Still, I worked from my prior Reinforced Security Agent (PoC) experience (portfolio write-up) and laid out my underlying beliefs, like Ground-Truth.

3) Job Interview (Dept-Head Interview)

A video interview with the hiring manager (department head), covering the Work Sampling answer plus my job experience, knowledge, and how I'd apply it. Since my convictions are firm and my experience was (by that role's standard) a bit atypical, the interviewers seemed to sense that "his angle is a little different." They probed much deeper than my PoC level — an area where I know the concepts but lack the hands-on depth to answer with full confidence.

Even so, I tried to hold my ground throughout. What stuck with me most: even though the interviewers never told me, my guess — "do you happen to have this kind of system internally?" — turned out to match their actual system pretty well. I basically predicted an internal structure I was never told about and got it right. I wanted to show that I grasp the overall picture and can draw the big-picture architecture, and I think I did.

Result & Reflection

In the end, despite all that, my strengths were probably a bit weaker for an experienced-hire seat — the fit just wasn't a perfect match. It was a good experience, but the information you can glean from the job posting and the phone interview is only the tip of the iceberg, which made it tough. (When I was rejected, the email came within just two days.)

It was a bittersweet interview, but I learned a lot. If the chance comes again, I'll give it another shot.

For reference, my review of Samsung's new-grad coding test is here. ↓

[New Grad] Samsung Electronics Coding Test Review (Samsung Research / Innovation Center)

5 attempts at the Samsung Research coding test — application timelines, exam environment, language restrictions, and lessons from each round.

taystudios.com/blog