astronomer-cosmos

Finance data & AI portfolio: tested dbt models (variance + GPU cost attribution) on DuckDB, Airflow/Cosmos orchestration, a guardrailed Anthropic-SDK variance agent, and a live case-study site. CI-validated, synthetic data.

Cloud-native analytics warehouse — dbt on BigQuery thelook_ecommerce + Airflow (Astronomer Cosmos) orchestration + GitHub Actions CI. Kimball star schema, 21 models, 88 tests, 2 DAGs.

Este projeto implementa um Data Warehouse (DW) completo do zero, usando uma stack moderna de engenharia de dados. O domínio de dados escolhido é a análise de atrasos em voos comerciais nos EUA, com cerca de 318 mil registros reais.

End-to-end data platform: multi-broker ETL pipeline (CoinGecko, US Treasury, GitHub Events) → PostgreSQL Medallion DWH (Bronze/Silver/Gold) → dbt models via Astronomer Cosmos → FastAPI read layer → Metabase dashboards. Orchestrated with Apache Airflow 3, asset-based cascade triggering.

Pipeline de données orchestrée avec Apache Airflow (via Astro CLI) et Astronomer Cosmos, qui exécute un projet dbt sur BigQuery.

Fork with 2 merged upstream PRs: #2201 (Watcher sensor template_fields fix) and #2241 (restore plain-text dbt logs in Watcher mode).

멀티테넌트 센서 데이터 ELT 파이프라인 — dbt + Airflow(Cosmos), PostgreSQL/TimescaleDB 기반. 증분 staging→mart 집계 모델, 시간 주기 데이터 품질 검증, 테넌트별 동적 DAG 생성.

Built an end-to-end ELT pipeline using Apache Airflow, DBT, and Cosmos to transform e-commerce data from MySQL into analytics-ready data marts in PostgreSQL. Implemented a layered Bronze, Silver, and Gold architecture with incremental loading, reusable DBT macros, hooks, Jinja templating, and SCD Type 2.

End-to-end ELT lakehouse pipeline: PySpark/Polars ingestion → dbt medallion modeling → Airflow + Astronomer Cosmos orchestration → Delta Lake

End-to-end modern ELT and analytics engineering project: Fivetran → Snowflake → dbt → Airflow + Astronomer Cosmos, validated with GitHub Actions CI.