Hands-on Data & AI projects, built and shipped.
A monthly HR analytics pipeline on Azure, built as an ELT lakehouse. A Python step lands full-roster snapshots in a raw zone (ADLS Gen2); a Synapse serverless SQL transform cleans and models them into a curated parquet layer through a reusable view and a CETAS stored procedure. GitHub Actions orchestrates the monthly run and authenticates to Azure as a least-privilege service principal; Power BI reads a consolidated serving view. Terraform provisions the lake, Synapse, and the RBAC.
A production-grade analytics engineering pipeline. HubSpot CRM and a billing source flow through Snowflake and dbt into 7 marts (MRR, ARR, funnel, account health). GitHub Actions runs the full pipeline daily; schema drift detection catches HubSpot changes before they break extraction; Reverse ETL pushes computed metrics back into HubSpot, closing the loop between analytics and operations.
A RAG chatbot grounded in Lucas's professional background, including his resume, VEGA case studies, and Coca-Cola Analytics Engineering work. Ask anything about his career, projects, or skills.
A multi-agent HR system over three MCP servers (HR data, jurisdiction rules engine, policy RAG). HR Business Partners ask in plain English: process a termination, convert a contractor, look up severance. The agent cross-references all three sources before acting. Compliant requests auto-execute and are audit-logged. Non-compliant ones are blocked and turned into a structured brief for HR review with the conflict, risk level, and recommended action.
An agentic chat over a synthetic SaaS dataset. The agent picks its own tools (cohort retention, funnel conversion, anomaly detection, industry benchmarks), writes SQL when needed, and renders interactive Plotly charts inline. Built directly on the Anthropic SDK with no agent framework.