17+ years in data. Five industries, banking to healthcare.
Senior Data & AI Engineer. Eight years building cloud data platforms — lakehouse on Databricks and AWS, warehouse modeling in Snowflake and dbt, multi-agent systems on Bedrock — shipped inside regulated environments.
Eight roles — seven employers plus my own practice, in reverse order. Every entry maps to a LinkedIn role and a stack you can dig into on a call.
Senior Data & AI Agent Engineer. Independent contractor providing data engineering and AI agent development services to clients in regulated industries. Work involves LLM-based observability, multi-agent orchestration, retrieval-augmented systems, and VPC-private model inference for environments with data-residency or compliance requirements.
Engagements covered under NDA — architecture and outcomes discussable on a call.
Data Engineer. Cross-cloud ingestion from REST API and Kafka sources into Databricks, Delta Lake, Unity Catalog, Snowflake and Azure SQL — with data contracts enforced as CI/CD gates in GitLab on every pull request. Databricks Workflows and Airflow on Kubernetes with per-DAG images to cut cold-start overhead. Redshift cost down via sort/dist key redesign, materialized dbt models and query rewrites. Built a LangChain + Claude assistant that parsed pipeline logs into ranked root causes for on-call review.
AWS Data Engineer. Redshift analytical performance improved through dimensional-model redesign, materialized views and workload-management tuning. Pipelines converted into parameterized factories with GitLab CI/CD validation, so any team member could ship a change through review.
AWS BI Data Engineer. Led the migration off SQL Server and SSIS onto AWS, rebuilding ETL on Glue, Lambda and Step Functions and serving analytics through Redshift, Athena and QuickSight. Cost cut by tuning Glue worker types, enabling job bookmarks for incremental loads and rescheduling production windows.
Senior Data Engineer. Productionized data-science models: notebooks refactored into production Python, inference on Lambda + Glue, version control, unit testing and CloudWatch/EventBridge monitoring discipline. Large-scale ETL on AWS (EMR, Athena, S3, Redshift, Kinesis).
Data Scientist Project Leader. Owned roadmap and delivery of Big Data & Analytics projects for Argentina's Visa acquirer across fraud, risk and merchant analytics. S3 · Athena · PySpark · Docker.
Data Analyst → Senior Data Scientist (Marketing & Credit Risk). Propensity, cross-sell and segmentation models across insurance and lending; a lending qualification engine, a real-time recommender on Oracle, and an NLP ReMarketing chatbot. Measurable sales lift, material call-center savings.
Case study overview →Data Analyst, Credit Risk Management. Seven years building and maintaining credit scoring models for retail and corporate portfolios — consumer loans, cards and SME credit lines. Python · SPSS · SQL.
Where the data comes from, what gets built, what ships.
Airflow task logs, cloud alarms, dbt test failures, on-call Slack noise.
Multi-agent triage over normalized events: correlation, severity scoring, runbook retrieval via RAG — inference kept inside client infrastructure.
Triaged notification with root-cause hint, linked runbook, next action. Fewer pings, faster MTTR.
Postgres, S3 dumps, SaaS APIs, Kafka streams. Mixed schemas, mixed cadences, no shared dictionary.
Modular ETL/ELT on AWS or Azure, Spark on EMR/Databricks for heavy load, dbt for modeling and tests. Cost and partitioning tuned from day one.
Curated tables in Redshift or Snowflake. BI-ready, SLA-tracked, documented. Storage and compute cut by a third.
Transactional warehouse, CRM attributes, behavioral signals, campaign history.
Feature engineering plus classical models (logistic, GBM, RFM, unsupervised), or embeddings and RAG when text dominates. MLflow tracking, batch inference in production.
Scored customers piped to CRM, campaigns, credit decisions and churn watchlists. Decisions move from gut to evidence.
Not a laundry list. Tools shipped to production, grouped by where they sit in the data path. Highlighted = primary in the last two years.
+ 4 more in databricks-examples · dbt-examples (ML/MLflow, cost optimization, testing, performance)
A 30-minute call to see if there's a fit. No pitch.
Defined scope and timeline. Typically 4–12 weeks.
Senior part-time capacity, embedded in your team. Typically 2–3 days a week.
Architecture, code review, technical decisions. Monthly retainer — async plus a weekly call.
Buenos Aires (UTC−3) · 6+ hours overlap with US Eastern · Contracting through Napoli Data LLC · I reply within one business day
Third-party proof: read the recommendations on LinkedIn →