Sample document — fictional candidate for illustration purposes only. Your CV is prepared from your own record.
Professional Summary
Data Engineer with 3 years of experience building high-throughput ELT pipelines and cloud data platforms at Kakao Corp and Wise. Specialises in Spark-based batch and streaming architectures, Snowflake data modelling, and infrastructure-as-code with Terraform. Comfortable owning pipelines end-to-end — from ingestion to analyst-ready data models — in environments where reliability is non-negotiable.
Pipeline & Platform Metrics
2.4 bn
Events processed per day
99.96%
Pipeline uptime over 18 months
55%
Snowflake compute cost reduction
8 min
Avg end-to-end latency (was 3 hrs)
Technology Stack
Apache Spark
Apache Airflow
Snowflake
Terraform
Python
SQL
dbt
Kafka
AWS (S3, Glue, EMR)
GCP (BigQuery, Pub/Sub)
Docker
Kubernetes
Scala
Git
Experience
Wise, London
- Business question: Could the monolithic nightly batch job be replaced to give finance and compliance teams intraday data? Approach: Redesigned the ELT architecture using Kafka + Spark Structured Streaming into Snowflake; rewrote 60 Airflow DAGs to micro-batch with 15-minute windows. Outcome: End-to-end data latency dropped from 3 hours to 8 minutes, enabling same-day FX reconciliation for the finance team.
- Business question: Was Snowflake compute spend growing disproportionate to data volume? Approach: Audited warehouse usage, identified 12 large scans caused by missing clustering keys; implemented automated clustering and query result caching policies. Outcome: Monthly compute cost reduced 55%, saving approximately £18,000/month.
- Built and open-sourced an internal Terraform module for Snowflake RBAC provisioning; adopted by 3 squads within 2 months.
Kakao Corp, Seoul
- Business question: How could the KakaoTalk analytics team ingest 2.4 billion daily events without pipeline failures during peak hours? Approach: Rebuilt the ingestion layer with Kafka partitioned by user region and a Spark job on EMR with auto-scaling; added dead-letter queue handling and alerting. Outcome: Pipeline achieved 99.96% uptime over 18 months with zero data-loss incidents during 3 peak promotional events.
- Business question: Could data engineers self-provision infrastructure without raising tickets to the cloud ops team? Approach: Authored Terraform modules and CI/CD pipelines (GitHub Actions) covering EMR, Glue, and S3 lifecycle policies. Outcome: Provisioning time for new data pipelines fell from 5 days to 4 hours.
- Mentored 2 junior engineers; co-led the data engineering guild (12 members) driving Python coding standards across the team.
Kakao Corp, Seoul
- Business question: Were legacy PySpark scripts slowing down the weekly reporting batch? Approach: Profiled job execution plans, replaced 8 wide shuffles with broadcast joins and partition pruning. Outcome: Weekly batch runtime reduced from 4.5 hours to 1.1 hours, freeing cluster capacity for additional workloads.
- Contributed to the migration of 35 on-premise Hadoop jobs to AWS EMR; all migrated jobs passed acceptance testing within schedule.
Additional Skills
- Data modelling (Kimball, Data Vault)
- Stream processing architecture
- Cost optimisation & FinOps
- CI/CD for data pipelines
- Data contracts & schema governance
- Korean (native), English (fluent)
Education
BSc Computer Science
Mar 2017 – Feb 2021
Korea Advanced Institute of Science and Technology (KAIST) — Top 10% of cohort, Specialisation in Systems & Distributed Computing