Shopee AI Engineer interviews, as three candidates describe them, mix LLM-focused technical discussion (RAG, LoRA, transformers) with coding screens and several stakeholder rounds. See what to prepare.
$155K
Avg. Base Comp
$190K
Avg. Total Comp
Not reported
Typical Rounds
Not reported
Process Length
Three candidates describe Shopee's AI Engineer process, and all three report no offer. Their accounts differ in format, so treat them as three separate experiences rather than a fixed script.
The most consistent technical theme is depth on modern LLM topics. One candidate was asked to go deep on RAG techniques, then on LoRA fine-tuning, reinforcement learning approaches and how training pipelines would be structured. The conversation felt exploratory rather than a live coding session, but concrete knowledge was expected. Two other candidates were asked to explain the transformer in detail rather than just naming it, and one of them was expected to cover the architecture and attention mechanisms.
One candidate describes a more traditional screen. It was a video call with roughly thirty minutes for a LeetCode-style problem, followed by resume discussion, LLM and machine learning fundamentals, and data sampling methods. A second technical round repeated that format with deeper concept questions. That candidate then met HR, an area coordinator, an operational manager and an area manager, and was asked about a project that went badly. Another candidate was told to expect three technical assessments covering NLP algorithms, machine learning models and programming, then team interviews and a senior leader review.
Two candidates mention communication problems or a lengthy process, so expect possible gaps between stages. Prepare to explain your own LLM projects end to end, including why you chose a given architecture, how you would integrate a model into a production workflow, and what you learned from failures.
Synthesized from 3 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Shopee process.
The process at Shopee for the AI Engineer role was longer than I expected, with multiple rounds spanning several weeks. My first round was a technical screening conducted over video. I had about thirty minutes to solve a LeetCode-style coding problem, which felt rushed but manageable. After that, the interviewer pivoted to discussing my resume and asking about my understanding of LLMs, machine learning fundamentals, and data sampling methods. The questions felt designed to gauge both my practical coding ability and my conceptual grasp of the field.
I made it through to a second technical round, which followed a similar format. There was another coding question and more resume-based discussion, but this time they dug deeper into specific concepts. They asked me to explain transformers and how they work—not just the high-level idea, but the mechanics. It was the kind of question where they clearly wanted to see if I could articulate the architecture, attention mechanisms, and why they've become so foundational in modern AI.
Beyond the technical rounds, the process included several interview stages with different stakeholders: HR, the area coordinator, an operational manager, and finally the area manager. Each seemed to serve a different purpose in evaluating fit. The behavioral aspects were mixed in throughout—one interviewer asked me about a project that didn't go well and how I handled the disappointment. These conversations felt more about understanding how I work with teams and handle setbacks rather than testing my technical knowledge. Overall, it was a comprehensive evaluation, though the process was lengthy and the technical bar seemed high.
Prep tip from this candidate
Go in with a strong understanding of transformer architecture and how LLMs work at a conceptual level, not just surface-level knowledge. Practice LeetCode problems under time pressure, especially in the first 30 minutes of a technical screening, and be ready to explain your resume projects in detail and discuss lessons from projects that failed.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Shopee
Describing a data project and its challenges
| Question | |
|---|---|
| Real-Time Transaction Streaming | |
| Categorize Sales | |
| Find Duplicate Numbers in a List | |
| Messenger Service Design | |
| Optimize Model Performance | |
| Open Source Reporting Pipeline | |
| Youtube Recommendations | |
| Weighted Average Sales | |
| LRU Cache 1 | |
| Why Do You Want to Work With Us | |
| Your Strengths and Weaknesses | |
| Unified Live Comments | |
| Regularization and Validation | |
| Empty Neighborhoods | |
| 2nd Highest Salary | |
| Rolling Bank Transactions | |
| Top Three Salaries | |
| Customer Orders | |
| Comments Histogram | |
| Closest SAT Scores | |
| Employee Salaries | |
| Merge Sorted Lists | |
| Subscription Overlap | |
| Size of Joins | |
| Compute Deviation | |
| Download Facts | |
| Monthly Customer Report | |
| Paired Products | |
| Upsell Transactions |
Synthesized from candidate reports. Individual experiences may vary.
Candidates report starting with a recruiter contact or a resume review and a short screen to gauge relevant experience. One candidate was contacted on LinkedIn and then waited a long time for the next step, so delays between stages may happen.
One candidate describes a video screen with about thirty minutes for a LeetCode-style problem, then resume discussion, LLM basics and data sampling. Another was told to expect technical assessments on NLP, machine learning and programming. Formats appear to vary by candidate.
Candidates report detailed questions on RAG, LoRA fine-tuning, reinforcement learning approaches, training pipelines and how the transformer works, including attention. One candidate described the discussion as exploratory rather than a live coding exercise, with high expected depth.
One candidate reports a second technical round with another coding question and deeper resume-based discussion, including requests to explain transformers mechanically. Expect to defend design choices and project outcomes in detail.
One candidate reports meeting HR, an area coordinator, an operational manager and an area manager, with behavioral questions such as a project that went poorly. Another was told the process ends with a senior leader evaluation. These stages may differ by team.
