About

I’m a data scientist graduating from the University of Southern California (USC) with an M.S. in Applied Data Science in December 2026. I work in Python and SQL, build data pipelines and machine learning models, and evaluate LLMs.

Recent work

  • At PlayStation, I built an LLM-assisted entity-matching pipeline for 186 organizations.
  • At Keck Medicine of USC, I built semantic models and machine learning forecasts.

Domain knowledge

  • Healthcare: healthcare BI, predictive operations, healthcare financial analytics
  • Cybersecurity: threat intelligence, vulnerability management
  • Marketing: ROI analytics, digital channels and growth, product strategy and market research

Interests: Human–AI trust, calibration, evaluation, fairness, and reliable AI systems.

Languages: Portuguese (native), English and Spanish (fluent)

Experience

Sony Interactive Entertainment (PlayStation) · Data Science Intern, Global Vulnerability Management

May 2026 – Present

  • Built an entity-matching framework that combines AI agents, probabilistic scoring, and deterministic rules to resolve machine identities across security tools (Qualys, CrowdStrike, Wiz)
  • Designed it as an hourly Snowflake pipeline sized for 800M+ records per run, giving 186 organizations one stable ID per machine and an 80-column asset inventory
  • Tested the LLM matching step like a classifier and reached 95%+ accuracy on calibration (two models vote, a third breaks ties)
  • Made every run auditable and reversible with 278 versioned rules, 56 data quality gates, SCD2 lineage, and anomaly alerts

Keck Medicine of USC

Data Science Intern

Apr 2026 – Present

  • Co-designed a time-series forecast of ICU bed capacity so a pilot team can plan beds and speed up discharges
  • Built a no-show risk model (logistic regression, XGBoost) and tuned its thresholds so clinics know when to double-book
  • Designed an automated AWS machine learning pipeline with leakage-safe features and per-clinic model calibration

Analytics Engineer Intern

Mar 2025 – Apr 2026

  • Raised Power BI Copilot answer accuracy 35% by building a 100+ question golden set to test and tune it
  • Co-designed one shared Power BI semantic model (SSAS Tabular, 150+ DAX measures) that replaced legacy SSRS finance reports

Diagnostics of America · Analytics Engineer

Jun 2021 – Dec 2022

  • Merged scattered finance and operations data into one GCP database (Alteryx, Databricks) for 700+ clinics and labs
  • Rebuilt company-wide FP&A reports on that database, cutting the monthly cycle from 4–6 days to ~5 hours

Decathlon Brasil · Marketing Analytics Intern

Dec 2019 – Dec 2020

  • Raised ad ROI 29% by analyzing Instagram and Google Ads spend and moving budget to what worked
  • Built Google Analytics reports on traffic and conversions, and combined them with market research to guide digital campaign strategy

Projects

DischargeAgent: Multi-Agent Discharge Coordination

Winner, Keck Hackathon (2026). A six-agent LangGraph system with a capacity-forecasting agent; targets 15%+ more early discharges and 25% fewer delays.

LangGraphReactFastAPIAWS Bedrock AgentCore
Featured Project

Document Intelligence

Winner, Keck Hackathon (2025). An eight-node LangGraph pipeline that turns scanned clinical documents into schema-validated JSON while keeping PHI on premises. Combines OCR, constrained LLM prompting, and append-only schema evolution to deliver consistent data for analytics, dashboards, and downstream ML models.

PythonFastAPILangGraphPostgreSQL (JSONB)Docling OCRReactTypeScriptOpenAI APISSEJSON Schema

NLP Q&A over BI Semantic Model (Keck Medicine of USC)

Q&A layer on a Power BI / SSAS Tabular semantic model that exposes KPI metadata, synonyms, and lineage to Copilot; raised Copilot answer accuracy 35% on a 100+ question golden set.

Power BISSASDAXCopilotSQL Server

Federated Healthcare Data Lake & Warehouse

Merged scattered finance and operations data from 700+ clinics and labs into one GCP database (Alteryx, Databricks), then rebuilt company-wide FP&A reports on it, cutting the monthly cycle from 4–6 days to ~5 hours.

GCPDatabricksAlteryxSQLETL/ELT

Research

Trust Orbit Model: Human Trust in AI

Research Assistant · USC Institute for Creative Technologies · PI: Dr. Gale Lucas · 2025–2026

Method:
A qualitative study with 113 participants, analyzed with thematic analysis.
Finding:
Trust Orbit Model. It names six cues that move a person’s trust in AI tools up or down.

Skills

Modeling & Statistics

  • Python (pandas, NumPy)
  • scikit-learn
  • XGBoost
  • Time-series forecasting
  • Anomaly detection
  • Logistic regression
  • Survival analysis
  • Statistical modeling
  • Probabilistic record linkage
  • Entity resolution
  • Model calibration

ML & AI

  • MLflow
  • LLMs
  • LLM-as-judge evaluation
  • LLM evaluation with golden datasets
  • Responsible AI auditing (Fairlearn, bias & data-leakage detection)
  • LangGraph
  • LangChain
  • RAG
  • Hybrid retrieval (BM25, Pinecone)
  • Prompt/context engineering
  • AI risk assessment
  • Bias mitigation
  • Agent orchestration

Data & Cloud

  • SQL
  • dbt
  • Data modeling
  • SCD2
  • CDC (change data capture)
  • Data quality & alerting
  • Snowflake (Tasks, Streams, Time Travel, Cortex AI)
  • Databricks
  • AWS (S3, Redshift, Lambda, SageMaker)
  • GCP
  • Alteryx
  • Streamlit
  • Security tools (Qualys, CrowdStrike, Wiz)

BI

  • Power BI dashboards
  • DAX & SSAS Tabular
  • Power BI Copilot semantic models
  • Data visualization
  • Microsoft Fabric
  • Tableau

Education

University of Southern California · M.S., Applied Data Science

Aug 2024 – Dec 2026 (expected)

Coursework

  • Natural Language Processing
  • Machine Learning
  • Information Security
  • Research Methods for User Studies
  • Database Systems
  • Data Mining
  • Foundations of Data Management
  • Fairness in AI

Activities

  • Research Assistant
  • Member, GRIDS (Graduates Rising in Data Science)

Conferences

  • Grace Hopper Celebration
  • AI Conference 2025 & 2026
  • Databricks Data + AI Summit
  • MCPCon

University of California, Irvine · Professional Certificate in Data Science

Jan 2023 – Aug 2024

Coursework

  • Calculus II
  • Calculus III
  • Linear Algebra
  • English for Academic Research
  • Data Science Professional Certificate

FIAP (Brazil) · B.S., Business Analytics

Jan 2019 – Dec 2022

Coursework

  • Business Intelligence
  • SQL
  • Statistics
  • Database Systems
  • Data Governance
  • Python
  • Data Visualization
  • Machine Learning
  • Big Data
  • Market Research
  • Consumer Behavior

Contact

Looking for full-time data science and applied AI roles across the U.S., starting January 2027. Hiring, or want to talk data and AI? Email me.

Email meLinkedInDownload resume