# Kmong Data Engineer — Remember Posting 334810
AI Summary
Purpose:
- Preserve the 2026-08-20 capture of Kmong's
Data Engineer (3+ years)posting and the evidence-based application decision.
Key points:
- The role combines data infrastructure and pipelines with production ML training/serving for recommendations, ranking, advertising, and LLM products.
- Verified candidate strengths are Python, Airflow, Kubernetes, data extraction/modeling, MySQL-centered data infrastructure, AWS EC2 operations, and an LLM API-based data product.
- Direct production evidence is missing for Spark, formal data-warehouse operation, recommendation/ranking model training and serving, feature stores, SageMaker/Amazon Personalize, Terraform, and real-time streaming.
- Estimated evidence coverage is about 60%. This is a stretch application, not a priority application: a low-cost application is reasonable, but a full tailored-resume effort should not displace higher-fit data-platform or database-platform roles.
- The posting showed a 2026-08-30 deadline, Seoul Seocho-gu location, negotiated compensation, a bachelor's-degree minimum, and a 35-hour work week.
Relevant when:
- Deciding whether to apply to Kmong.
- Tailoring a future Kmong resume or preparing screening answers about ML serving, data warehousing, Spark, and AWS scope.
Do not read full document unless:
- The exact requirement breakdown or evidence gaps are needed.
Linked documents:
../../wiki/projects/2026-active-job-shortlist.md../../wiki/projects/2026-career-transition.md../../wiki/people/hyunwook.md../../wiki/projects/data-platform-systems-engineering.md../../wiki/projects/labrador-platform.md
Open Questions
- Is production recommendation/ranking model training and serving a strict screening gate, or can a strong data-platform engineer grow into it?
- What warehouse technology, Spark workload, orchestration scale, and on-call expectations does the team operate?
- What percentage of the role is data engineering versus ML platform/MLOps work?
- What is the compensation range?
- The current team size, reporting line, and interview emphasis are Unknown.
Details
Posting metadata
- Company: Kmong (
(주)크몽) - Role:
Data Engineer (3년 이상) - Experience: 3–7 years
- Education: bachelor's degree or higher
- Location: Seocho-gu, Seoul
- Compensation: negotiable
- Deadline shown on capture: 2026-08-30
- Process: document review → first interview → second interview → compensation negotiation → start
- Benefits explicitly listed: 35-hour work week, annual KRW 500,000 role-related education support, flexible leave, birthday half-day, lunch on office days, health examination support, long-service recognition, guild gatherings, and referral rewards.
Responsibilities
- Build and operate data infrastructure.
- Build and operate data pipelines.
- Develop and operate data products, including recommendation and advertising model training/serving and LLM products.
Required qualifications
- Practical use of data technologies such as Python, Spark, and Airflow.
- Practical use of AI technologies.
- Data modeling and extraction.
- Understanding and operation of AWS infrastructure.
- Production operation of training and serving pipelines for recommendation, ranking, or similar ML models.
- Three to seven years of data-engineering experience.
- Data-warehouse design and operation.
- Data-infrastructure construction and operation.
- Kubernetes service operation.
- Proactive problem discovery, purpose-aware decisions, and delegating repetitive work to AI.
Preferred qualifications
- Terraform.
- Amazon Personalize or SageMaker.
- Feature-store construction and operation.
- AI/LLM product development or operation.
- Data-product design and operation.
- Real-time infrastructure and pipelines.
- Recommendation-system or advertising-platform domain experience.
Evidence mapping
| Requirement | Evidence status | Verified candidate evidence |
|---|---|---|
| Python / Airflow | Strong | More than 80 scheduled crawlers and Airflow/Kubernetes pipeline operation |
| Spark | Gap | No verified production Spark experience |
| Practical AI use | Partial-to-strong | ChatGPT API-based license classification; earlier k-NN research |
| Data modeling and extraction | Strong | Multi-source vulnerability/license collection, schema design, row-grain and conflict-rule design |
| AWS infrastructure | Partial | Self-managed workloads on EC2 only; no verified RDS, S3, or EFS production use |
| Recommendation/ranking ML training and serving | Critical gap | No verified production evidence |
| Three to seven years of data engineering | Strong | LabradorLabs from 2021.06 to present |
| Data warehouse design and operation | Major gap | Strong operational MySQL/data-distribution work, but no verified formal DW or analytical-mart operation |
| Data infrastructure | Strong | Hybrid IDC/in-house/AWS EC2 architecture, DB distribution, backup, monitoring, and incident response |
| Kubernetes operation | Strong | Six-node topology work, Airflow workloads, capacity and networking operation |
| Terraform | Gap | No verified production evidence |
| Personalize / SageMaker | Gap | No verified production evidence |
| Feature Store | Gap | No verified production evidence |
| AI/LLM product | Partial-to-strong | Production-oriented LLM API integration for license-data classification |
| Data product | Strong | Vulnerability/license collection and customer data-delivery systems |
| Real-time pipeline | Gap | Verified CDC is intentionally batch-oriented; do not present it as streaming |
| Recommendation/advertising domain | Gap | No verified production evidence |
Decision
- Estimated evidence coverage: approximately 60%, with uncertainty because the posting does not reveal how strictly the ML-serving requirement is screened.
- Recommendation: optional low-cost stretch application before 2026-08-30.
- Do not spend priority resume effort unless the user deliberately wants an ML-platform transition.
- If applying, lead with Airflow/Kubernetes, data-source modeling, data quality, AWS EC2 operations, and the LLM data-product example. State the ML-serving, Spark, and warehouse gaps honestly.