Romaric Thiam

Team Lead Data Engineering @Kadmos

Berlin, DE

Building and managing the team- Hired and managed the Data Team: a Data Analyst and a Data engineer. Mentor team members on technical issues, conduct code and design reviews, and enforce Data-Engineering best practices- Maintain sync with BackEnd, FrontEnd and Product team. Acting as the Data Team Product Owner. Building a Data infrastructure from scratch on the Google Cloud Platform (GCP)- Data warehouse on Google Cloud Storage (GCS)+ BigQuery. Building a standard Data-Lake/Data-Mart warehouse- PySpark ETLs. Airflow as a scheduler to build a DataWarehouse with hundreds of tables- Metabase as an internal Data-Visualisation tool- ”Reverse ETLs” replicating parts of the Data Warehouse in Back End services (Postgres, Spring Boot) to expose datasets to external customers- ML training pipelines (SparkML) for building logistic regressions used for forecasting- Internal LLM service- All Data services dockerised, deployed with Kubernetes (Github CI, Bash, Docker)- Data Engineering best practices: Pre-aggregated reports, compression, partitioning, data governance, “functional data engineering” patterns. Despite having dozens of thousands of queries on Metabase per month, we kept BigQuery Queries cost under a few dozens euros per month.Building business value- ”Self serve” BI approach. Building reports for Finance, Ops, Founders’ office, Product teams and customers. Most of the company regularly access Metabase with minimal support rather than relying on an extensive Data-Analyst team- High level business insight reports are fully automated, by redirecting Reporting ETL outputs to our internal LLM service- Ops automation/Workforce management, the Data Team is the main proactive trigger of Customer Support work. All business-critical workflows are covered by automated alerts that proactively notify customers, and in some cases provide LLM-generated responses to assist the support team- ”Report builder” for external corporate customers.