Bruno – Python, SQL, AWS

Worked on the development and optimization of advanced data and AI pipelines to support large-scale content processing and analytics for a leading media organization, improving the efficiency and reliability of data flows and enabling better insights from complex datasets.

•   Build and maintain scalable data pipelines on AWS MWAA (Apache Airflow), AWS S3, Snowflake, Databricks, and dbt, turning raw content data into self-serve datasets that newsroom analytics and editorial teams rely on.

•   Designed and shipped a production semantic layer from scratch on Snowflake and dbt. It models content, engagement, and editorial metrics as governed entities with reusable metric definitions, used across editorial, product, and exec stakeholders.

•   Built LLM-based agents on top of OpenAI APIs and tool-calling frameworks (text-to-SQL and semantic-layer-backed Q&A), with guardrails against hallucinated metrics and answer-quality evals so accuracy stays auditable.

•   Set up data-accuracy controls from day one: reconciliation against source-of-truth APIs and PostgreSQL/MongoDB operational stores, schema contracts, referential-integrity tests, cardinality assertions, and daily drift checks, all wired into CI/CD.

•   Optimized SQL and reshaped data models on Snowflake and BigQuery (clustering, partitioning, materialization strategy, query rewrites), with deliberate judgment on when rewriting SQL was enough versus when the model itself had to change. Cut both latency and warehouse cost.

•   Wrote column-level data dictionaries, runbooks, and onboarding guides so the team could scale documentation practices and reduce on-call load.

A project focused on developing and improving data marts for a personal finance platform, aiming to enhance data organization, accessibility, and reliability to support analytics, reporting, and business decision-making.

•   Delivered 4 Data Marts and 30+ dbt models (tables, views, macros) on Snowflake and Databricks, orchestrated with Airflow. The work turned exec questions about MRR, churn cohorts, and customer funnels into self-serve data products.

•   Designed semantic-layer entities and metric definitions from scratch (not inherited and patched), so business teams could query governed metrics without re-deriving the same logic in every dashboard.

•   Ran query-optimization passes on Snowflake and BigQuery (clustering keys, materialization strategy, SQL rewrites). Query times dropped and real-time data became more accessible to the business.

•   Built reconciliation jobs against Segment source-of-truth events and PostgreSQL operational replicas, plus referential-integrity, uniqueness, and cardinality tests in dbt. Caught accuracy incidents before they reached executives.

•   Partnered with BizOps and Finance stakeholders to model SaaS metrics end-to-end (MRR, ARR, retention, expansion), turning ambiguous business questions into durable, well-tested data products.

•   Wrote detailed documentation (data dictionaries, lineage, runbooks) and tuned orchestration patterns to simplify workflows and reduce on-call load.

The project focused on building the first regional Data Warehouse for the LATAM market, starting with the Supply Chain domain. The goal was to centralize and streamline data from multiple sources, enabling better visibility into logistics, inventory, and operational performance across countries in the region. This foundation later supported the expansion of data analytics and reporting capabilities for other business areas.

Saint-Gobain: •   Ran the data warehouse and built 20+ models (tables, procedures, functions, tasks, views) on Snowflake, BigQuery, dbt, Databricks, and Azure for web, targeting, and supply-chain analytics.

•   Saint-Gobain: led the SAP-to-Snowflake migration with intermediate staging on PostgreSQL, designed the ETL pipelines, applied reconciliation and integrity checks against source SAP extracts, and shipped Power BI dashboards covering supply-chain KPIs.

Globo: •. Ran audience-segmentation analytics on BigQuery with Google Ad Manager data, built Looker Studio dashboards used by marketing and content strategy teams, and modeled funnel and cohort metrics for executives.

•   Tuned BigQuery and Snowflake query performance with partitioning, clustering, and SQL rewrites to balance cost and latency on production reporting workloads.

•   Documented column-level metric definitions and lineage so reporting was consistent across business teams.