Grigorios Konstantinos Gkaravelos

I help businesses and data teams to replace manual Excel reporting and build BI-ready data pipelines and reliable reporting systems. After two years as a data analyst, I transitioned to analytics engineering because I became more interested in the data behind the dashboards and the layer between raw data and the final report. I designed and built a production-style food data lakehouse that solves a real problem. Open Food Facts is a public API with 3M+ product records but zero analyst-ready tables. Nutrichain Open Food Lakehouse is a portfolio project demonstrating my skills in analytics engineering. It turns that raw API into a trusted, queryable lakehouse on Databricks:

- Engineered an end-to-end batch ingestion pipeline in Python fetching paginated JSON from the Open Food Facts API, landing raw data into a Bronze Delta Lake table with run-level metadata, hash keys, and audit logging for full lineage traceability

- Orchestrated daily pipeline runs using Apache Airflow with task dependencies, retry logic, and Databricks Secrets for API key management — targeting ≥95% job success rate

- Transformed Bronze data through a Silver layer using dbt on Databricks: deduplication on business keys, type casting, null handling, and schema validation across 500K+ records from 170+ countries

- Modelled a Kimball-style star schema in the Gold layer with dbt (fact_product_nutrition + dim_brand, dim_category, dim_nutriscore, dim_country) connected directly to Power BI via Databricks SQL

Contracted to execute a critical master data integration project following a company acquisition, specifically handling the Greek and French markets:

- Tasked with identifying and resolving duplicate customer records between an 8,000-record acquired database and a 514,000-record enterprise database

- Engineered custom Excel macros and utilized advanced data functions to automate the matching of customer names, addresses, and IDs across both datasets

- Successfully mapped and deduplicated the required records, ensuring a clean migration into the parent company's master database

Selected for an intensive, hands-on data analytics and business intelligence training program, culminating in an end-to-end capstone engineering project:

- Designed and built a local data warehouse from an OLTP source database, structuring the data into a STAR schema for analysis

- Engineered SQL-based ETL pipelines to extract, transform, and load the data, applying quality checks for null handling and primary key validation

- Developed a multiclass classification machine learning model (Python, Scikit-learn) to forecast sales performance

- Built the final visualization layer using Power BI and DAX to present the findings

- Documented the whole pipeline providing clear code and demonstrating end-to-end ownership of the full data engineering lifecycle