Open to remote roles and freelance projects

Justin Olivo

Applied AI Engineer

I build agentic AI systems that ship as working apps: retrieval over trusted sources, multi-model routing, and agent workflows. My depth is healthcare, after nine years working the ER floor.

See my work Contact me
  • FocusAgents, RAG, multi-model routing
  • StackPython, LangGraph, FastAPI, ChromaDB, Docker
  • ShippedCardioBot, a live multi-model assistant
  • Domain depth9+ years in healthcare, U.S. Army veteran

How I build

I focus on AI systems that answer from real sources and run as working software, not demos that fall apart on the first hard question.

Right model for each task

I route work across Claude, GPT-4, and Gemini based on what each does best, instead of forcing one model to do everything.

Retrieval over guessing

I ground answers in a document corpus so users can trace where an answer came from.

Domain judgment

Nine years in a high-acuity setting taught me where tools fail under pressure, and I design with that in mind.

Case study: CardioBot

A multi-model RAG assistant that answers clinical questions, reads EKG images, and handles dose calculations. It shows the full pattern: retrieval, multimodal input, and routing in one working app.

83%reduction in pre-arrival decision support time
895passages indexed from the 2023 ESC ACS Guidelines

The problem

Decision support needs to be fast and tied to published guidance. A single general-purpose model is neither reliably grounded nor equally good at text, images, and math.

The approach

Guideline passages live in ChromaDB for retrieval. Claude Sonnet handles questions, Gemini Vision reads EKG images, and GPT-4 handles calculations. LangGraph coordinates the workflow.

User inputQuestion, image, or calculation
RoutingChooses the right model for the task
Claude SonnetQuestions with retrieval
Gemini VisionImage interpretation
GPT-4Calculations
ResponseGrounded in source passages
  • LangGraph
  • ChromaDB
  • Python
  • Streamlit
  • Claude
  • GPT-4
  • Gemini
Open the live demo View the code

Portfolio project built for demonstration. Not for clinical use.

Capabilities

Agents and orchestration

LangGraph and LangChain workflows, multi-model routing, tool use

Retrieval systems

Document ingestion, vector search with ChromaDB, grounded answers, evaluation

Build and deploy

Python, FastAPI, Streamlit, Docker, AWS, with an Azure deployment in progress

Model APIs

Anthropic, OpenAI, Google, and Hugging Face

More projects

  • CardioBot v2In progressNext version of the assistant, with deployment on Azure.
  • TriageBotPlannedAn agentic triage system, the next project in my portfolio.

Work with me

Custom AI assistants

Purpose-built AI agent assistants for individual professionals, scoped to the way you already work. Available through Upwork.

AI tool and workflow reviews

A practical review of the AI tools and workflows you use, with clear recommendations for what to keep, change, or drop.

Experience

  1. Now

    Freelance AI Engineer

    Building agentic AI projects and custom AI assistants.

  2. 2026

    Agentic AI Bootcamp, Codecademy

    Completed with certificate. Shipped CardioBot.

  3. 9+ years

    ER Technician, Level II trauma center

    Night shift in a high-acuity emergency department.

  4. 2013 to 2014

    U.S. Army, Health Care Specialist (68W)

    Army health care specialist training and service.

Certification: AWS Cloud Practitioner. Languages: English and Spanish.

Get in touch

I'm looking for remote Applied AI roles and take on select freelance projects. The fastest way to reach me is email, and LinkedIn works too.

justinolivo1207@gmail.com

Open in email app

LinkedIn justin-olivo52 GitHub JustinOlivo52 Upwork Hire me for AI assistants Contra Freelance profile