Raymundo Navarrete
Raymundo Navarrete
ML Engineer & AI Systems Architect

Building AI systems that ship — LLM pipelines, agentic workflows, and production ML. 10+ years across defense AI, FinTech, and autonomous systems. Targeting healthcare.

📍 Remote / Columbus, OH 🎓 PhD Applied Math · U of Michigan 🔐 Active Secret Clearance 🏥 Healthcare AI Target
📄 View Resume LinkedIn ↗ Email

About

I'm a machine learning engineer and data scientist with 10+ years of experience building models and systems that go to production. My work spans FinTech credit modeling, government-contracted defense analytics (Air Force supply chain, Secret clearance), and autonomous AI agent systems. I hold a PhD in Applied Mathematics from the University of Michigan and have a particular interest in where rigorous quantitative methods meet modern LLM tooling.

Currently targeting healthcare AI roles — I want to apply LLM orchestration, production ML pipelines, and systems-level thinking to problems that matter clinically. I build AI agents for fun: the system running this site is itself an autonomous overnight prototype builder powered by Claude.

Languages Python SQL R C/C++
ML / DL PyTorch scikit-learn xgboost HuggingFace survival models
LLM / AI Anthropic SDK LlamaIndex RAG / pgvector tool-calling Claude Code
Cloud AWS GCS Databricks Docker Kubernetes
Data PostgreSQL BigQuery Snowflake SQLite pandas / polars

Experience

Senior Data Scientist
Expansia Government Contracting Secret Clearance
Feb 2024 – Mar 2026
  • Led ML/AI architecture for an Air Force supply chain risk platform — improved detection of at-risk parts from 10% → 60% recall across 1M+ parts using xgboost, survival models, and PyTorch.
  • Built an agentic AI system (LlamaIndex + pgvector, AWS Bedrock, Anthropic SDK) allowing stakeholders to query risk scores and manufacturing candidacy in natural language.
  • Built OCR/NLP pipeline (Tesseract, AWS Textract) to digitize decades of scanned Air Force maintenance records into model-ready features; reduced manual part review burden by 90%.
xgboostsurvival modelsRAGLlamaIndexpgvectorAWS BedrockAnthropic SDKOCR/NLPDatabricks
Senior Data Scientist
Best Egg FinTech / Personal Loans
Mar 2024 – Sep 2025
  • Led end-to-end migration and automation of loan loss forecasting platform, delivering more reliable dashboards and reducing team workload by 2 days/month.
  • Designed alternative-data second-look model for a flexible rent product, increasing approvals while maintaining stable default rates.
credit modelingSnowflakeAWS MetaflowKubernetesforecasting
Senior Data Scientist
Battelle Science & Technology / Government Contracting
Mar 2023 – Feb 2024
  • Led procurement of $1M EPA contract and managed cross-functional team to deliver a predictive risk assessment model on time and within budget.
  • Applied LLM fine-tuning and prompt engineering to identify national security risks associated with chemical synthesis methods for adversary threat assessment.
LLM fine-tuningrisk modelinggovernmentAzureC++
Machine Learning Researcher
Upstart FinTech / Personal Loans
Apr 2021 – Feb 2023
  • Developed state-of-the-art ML techniques improving the flagship underwriting product, increasing quarterly performance gains by 100% and enabling the team to hit quarterly goals.
  • Conducted time-sensitive analysis to identify underperformance in personal-loan underwriting during economic turbulence; implemented corrections in production.
credit modelingscikit-learnxgboostPyTorchAWSAirflow
Postdoctoral Researcher
University of Arizona Academia
Aug 2018 – Mar 2021
  • Published multiple peer-reviewed papers on ML, graph neural networks, and dimensionality reduction in top journals (IEEE TVCG, Physica D, EJS).
graph neural networksML visualizationPythonresearch
PhD in Applied Mathematics
University of Michigan Ann Arbor, MI
2013 – 2018
  • Best applied mathematics thesis award. Research in dynamical systems, delay embeddings, and time series prediction using kernel methods.
dynamical systemsstatisticsnumerical analysis

Projects

Press / to focus filter — type skills, or

Publications

ENS-t-SNE: Embedding Neighborhoods Simultaneously t-SNE
Miller J., Huroyan V., Navarrete R., Hossain I., Kobourov S.
17th IEEE Pacific Visualization Symposium (PacificVis), 2024
dimensionality reductionvisualizationt-SNE
Multi-Perspective, Simultaneous Embedding
Hossain M.I., Huroyan V., Kobourov S., Navarrete R.
IEEE Transactions on Visualization and Computer Graphics, 2020
graph embeddingvisualization
Prevalence of delay embeddings with a fixed observation function
Navarrete R., Viswanath D.
Physica D: Nonlinear Phenomena, vol. 414, 2020
dynamical systemsapplied math
Delay embedding of periodic orbits using a fixed observation function
Navarrete R., Viswanath D.
Physica D: Nonlinear Phenomena, vol. 388, pp. 1–9, 2019
dynamical systemsapplied math
Prediction of dynamical time series using kernel based regression and smooth splines
Navarrete R., Viswanath D.
Electronic Journal of Statistics, vol. 12(2), p. 2217, 2018
time seriesstatisticsML
Accuracy and stability of inversion of power series
Navarrete R., Viswanath D.
IMA Journal of Numerical Analysis, vol. 36(1), pp. 421–436, 2016
numerical analysisapplied math

Let's Talk

Currently Seeking

Healthcare AI and senior ML/AI engineering roles — especially where LLM orchestration, production ML systems, and clinical impact intersect. Roles where I can own the technical architecture and drive real results.

  • Healthcare AI / clinical decision support
  • AI engineering / LLM pipeline design
  • Senior data science (ML systems in production)

Remote preferred · open to hybrid (Columbus, OH)

Reach Out
🔗 linkedin.com/in/raymundo-navarrete ✉️ [email protected] 📄 Resume PDF