Projects

Goal: Extracts invoice fields with Claude or Gemini, keeps the exact text behind every value, scores each field and sends doubtful documents to a person.

Tools: Python, Claude (structured outputs), Gemini, Pydantic, pytest, GitHub Actions

Impact: Shows why field accuracy alone misleads: a regex baseline gets 86% of fields right but only 25% of documents.

Goal: A dbt + DuckDB starter where Claude Code follows data rules, and hooks block DROP, production targets and secrets before they run.

Tools: Claude Code (hooks, subagents, skills), dbt, DuckDB, Python, pytest, GitHub Actions

Impact: 34 hook tests and 12 dbt data tests, including a reconciliation that fails if a join starts dropping orders.

Goal: Compares a legacy table with its migrated copy by hash, from day to row to cell, and filters out the harmless noise every migration adds.

Tools: Python, DuckDB, SQL, pytest, GitHub Actions

Impact: Compares 1,000,000 rows in 0.09s on a laptop and finds exactly the 3 planted bugs; exit codes make it a CI gate.

Goal: Three reproducible data incidents worked like investigations: symptom, hypotheses, evidence, root cause, fix and a guard that stays.

Tools: Python, DuckDB, SQL, pytest, GitHub Actions

Impact: Match rate restored from 57% to 98%, a 15% revenue overstatement closed, and 5 reusable guard checks.

Goal: Medallion-architecture data lake that ingests CRM, ERP and API data, uses Claude to advise on data quality, generate SQL and business insights, and exposes the lake to agents through an MCP server

Tools: Claude (Sonnet + Opus), MCP, AWS S3, Glue, Athena, PySpark, dbt, CloudFormation, pytest

Impact: End-to-end pipeline from raw data to business-ready KPIs, where AI agents query the lakehouse directly via MCP

Goal: Educational tutor running Gemma 4 locally on a community edge server, so primary students without internet can use AI from basic phones over local WiFi

Tools: Gemma 4, Ollama, Google ADK, FastAPI, Angular, RAG (TF-IDF, scikit-learn), Docker Compose

Impact: Built for GDG Lima's Build with Gemma competition (Local AI & Edge track): AI deployed where the user is, with no connectivity required

Goal: Production-grade RAG system with vector database, LLM integration, and automated evaluation using Ragas

Tools: ChromaDB, OpenAI/Groq, FastAPI, Ragas, Prometheus, Docker, CI/CD

Impact: Complete knowledge management pipeline with metrics tracking, performance monitoring, and model versioning

Goal: Spark-based lakehouse with Medallion Architecture (Bronze/Silver/Gold), streaming pipelines, ML training, and Airflow orchestration

Tools: PySpark, Delta Lake, MLflow, Airflow, Databricks, Terraform, Great Expectations

Impact: Scalable data platform processing millions of records with integrated ML pipelines, Delta Lake optimizations, and infrastructure provisioning

Goal: Calculate financial risk metrics (VaR, CVaR, Greeks) using Databricks with Airflow and Control-M orchestration

Tools: Databricks, Delta Lake, Control-M, Airflow, Great Expectations, PySpark

Impact: Enterprise-grade risk analysis system with data quality gates, Control-M integration, and CI/CD pipeline

Goal: Modern analytics engineering platform with dimensional modeling (Kimball), incremental models, and >80% test coverage

Tools: dbt, Databricks, Delta Lake, dbt_expectations, Z-ORDER, Liquid Clustering

Impact: Fully-tested analytics pipeline with customer segmentation (RFM), Delta Lake optimizations, and comprehensive documentation

Goal: End-to-end MLOps pipeline with MLflow tracking, FastAPI serving, drift detection (Evidently), and automated CI/CD

Tools: MLflow, FastAPI, Evidently AI, Prometheus, Docker, Great Expectations, GitHub Actions

Impact: Complete ML lifecycle system with model registry, drift monitoring, API serving, and automated deployment pipeline

Goal: Normalization, analysis and prediction of road accidents in Peru using 2008–2023 historical data

Tools: PySpark, Parquet, scikit-learn (Random Forest), Flask, Jupyter

Impact: ETL pipeline from Excel to Parquet, temporal and regional feature engineering, and a web app that predicts accidents by region, date and weather

Goal: Agent that finds Data Engineer roles daily, scores them against your CV and uses an LLM to write a tailored application pack per job

Tools: Python, JobSpy, SQLite, Groq / OpenAI, RenderCV, Jinja2, GitHub Actions

Impact: Automated daily run that deduplicates postings, filters out unrelated roles and outputs a tailored PDF CV, cover letter and recruiter message per job