Data Engineer · Modern Data Stack · Python · dbt · Spark · AWS · GCP
I build data platforms that are easier to trust, operate, and extend.
My background is in Java/Spark production systems on high-volume banking platforms. I now focus on Modern Data Stack architectures: Python ingestion, raw storage, dbt transformations, cloud warehouses, BI layers, data quality, orchestration, and clean handover to data teams.
| Problem | What I bring |
|---|---|
| Fragile Python or SQL ETL | Modular ELT design with clear responsibilities |
| Slow or risky batch pipelines | Incremental strategies, raw replay, validation, monitoring |
| Business logic spread everywhere | dbt layers, tests, documentation, governed marts |
| BI blocked by data debt | Analytics-ready models and self-service foundations |
| Cloud migration or platform rebuild | Practical architecture, delivery focus, team handover |
Serverless ELT platform on Google Cloud for Velib Metropole open data.
It captures the state of the bike network every 5 minutes, stores raw JSON in GCS, loads BigQuery, transforms with dbt, and exposes analytics-ready tables for a Streamlit dashboard.
| Area | Stack |
|---|---|
| Ingestion | Python 3.12, Cloud Run Jobs, Cloud Scheduler, Cloud Workflows |
| Storage & warehouse | Google Cloud Storage, BigQuery |
| Transformation | dbt-core, dbt-bigquery, incremental models |
| Infrastructure | Terraform, Docker, IAM least privilege |
| Quality | pytest, ruff, SQLFluff, CI/CD |
The goal is to show a complete data platform, not just a pipeline: reproducible infrastructure, raw replay, layered modeling, tests, documentation, and cost-conscious serverless execution.
| Context | Highlights |
|---|---|
| Fintech scale-up | Rebuilt a data platform in 3 months with Python, S3, Redshift Serverless, dbt, AWS Fargate, QuickSight, RBAC, and GDPR masking policies. Reduced a nightly pipeline from 3-4h to less than 1h. |
| Banking platform | Built and maintained Java/Spark distributed pipelines on production systems with Hive, SQL Server, Jenkins, and 24/7 operational constraints. |
| Banking group | Worked on Java, Spark, IBM ODM, business rules, distributed processing, TDD, and Agile delivery. |
| Data | Cloud | Engineering |
|---|---|---|
| Python | AWS S3 | Terraform |
| SQL | Redshift Serverless | Docker |
| dbt | AWS Fargate | GitHub Actions |
| Spark | QuickSight | Forgejo Actions |
| Hadoop / Hive | BigQuery | Jenkins |
| Data quality | GCS | pytest / ruff / SQLFluff |
- I prefer simple, explicit data flows over clever pipelines that nobody wants to touch.
- I separate ingestion, raw storage, transformations, marts, and BI responsibilities.
- I care about tests, documentation, naming, and operational clarity because data platforms live longer than their first delivery.
- I like working close to analysts, business teams, DevOps, infrastructure, and security.
Open to Data Engineering roles or freelance missions around platform modernization, ELT architecture, dbt, cloud warehouses, and reliable analytics foundations.
[email protected] · antoinegelin.alt144.net · github.com/gelinant