Christian Vega.

Python · SQL · Snowflake · Azure Data Factory · LLM integration

• Architected and shipped two in-house, LLM-powered automation products now in production across enterprise healthcare data — an automated Business Requirements Document (BRD) generator and a data-dictionary context generator — replacing documentation work previously produced by hand by business analysts.
• Owned each product end-to-end: scope, architecture, LLM and prompt design, output evaluation, and production rollout under HIPAA-governed data constraints.
• Led migration of legacy systems onto a modern, automated, ML/LLM-driven data platform, owning architecture and delivery.
• Set technical direction and coordinated cross-functional stakeholders on AI and automation initiatives, from design through production.

Python · FastAPI · PostgreSQL/pgvector (HNSW) · LightGBM · Nuxt 3 · TypeScript · Docker · AWS (Lightsail / ECS Fargate) · Cloudflare

• Solo-architected, shipped, and operate three live production systems end-to-end — data ingestion, ML, application, and production ops — using AI coding agents as an accelerated implementation layer under my own architecture, model-evaluation, and production-ops control.
• Nidopr (live at nidopr.app) — bilingual real-estate marketplace: ~64,000 listings ingested nightly from 10+ sources via resumable crawlers; a LightGBM valuation model with calibrated 80% prediction intervals (~79.6% empirical coverage) and a trust gate that cut model bias from +44% to +5% in back-testing.
• Nidopr production ops — 34 unattended nightly jobs and a 40-check monitoring layer alerting to phone, Discord, and email; an autonomous Facebook poster publishing 7 data-selected listings/day with OCR photo-safety screening.
• LicitaPR Inteligencia — RAG-backed procurement-intelligence product unifying six government data systems: ~1.24M contracts ingested (~90% of the universe) with Splink entity resolution (422K → ~345K entities); hybrid retrieval over pgvector/HNSW embeddings; kept green by a 62/62-check QA harness.
• FT-OS — complete small-business operating system (visual builder, POS, kitchen display, inventory, admin) live on AWS in daily production use, held green through a 148-check QA battery and git push-to-deploy.

MSSQL Server · SSIS · ETL · Python · Power BI · SSRS · REST APIs

• Built an automated ETL pipeline (SSIS + Python) consolidating data across multiple isolated silos — collapsing a manual compilation that previously took a business analyst ~2 months of full-time work, plus developer and multi-department support, into a single repeatable run.
• Designed and shipped end-to-end data pipelines and BI dashboards for PRASA, moving operational reporting from manual handling to automated, near-real-time delivery.
• Engineered REST API integrations to pull and process external sources into the warehouse, improving dashboard data coverage and accuracy.
• Automated reporting (MSSQL + Python + SSRS), cutting report-generation time ~30%.
• Led data-governance initiatives across PRASA analytics — integrity, security, and compliance.

SQL · Postman · Snowflake · SSIS · SSMS · Azure DevOps

• Built automated test frameworks (SQL + API validation via Postman) that cut manual testing time ~30%.
• Ran Snowflake data-model query testing and validation (SSIS) to certify accurate reporting structures.
• Planned deployments and test cases with cross-functional teams (~20% faster releases) and streamlined API testing to reduce unresolved bugs ~15%.

SQL · Power Automate · SSIS · IBM Cognos · Tableau

• Led compliance initiatives — automated audit workflows and data validation (SQL, Power Automate) — reaching a 95% compliance rate.
• Conducted data-integrity audits (SSIS, IBM Cognos) that reduced regulatory reporting errors ~10%, and drove ~25% faster claims/inquiry resolution via automated Tableau dashboards.
• Enforced HIPAA and Medicare Parts C & D adherence through data-governance policies and encryption.