I was trained as a neuroradiologist aiming to understand the brain, but I became an AI engineer experimenting with digital brains. I went into neuroradiology to figure out how the brain works, only to discover it doesn’t come with an instruction manual. Looking at its parts all day taught me about anatomy, but left me wondering how intelligence works. That question led me to AI research, where I could experiment with neural networks, and to engineering, where I became obsessed with automating my own work.
I helped build Cognita Imaging from founding to acquisition. What began as an effort to save myself some time led to medical AI research at Stanford and a place on the founding team at Cognita Imaging. We built AI to automate radiology workflows and, in less than a year, the company was acquired by Radiology Partners, the world’s largest radiology group.
At Hippocratic AI, I was responsible for the operations agentification and a new product called AI Orchestrator. At Hippocratic AI, I worked on automating company operations with agents and developed a new product called Agents Orchestrator.
I’m exploring what’s next at Outcast Ventures O2. I’ve since left Hippocratic AI to explore the next highest-leverage problem to solve with AI, now that we have AGI in our hands. I’m currently part of the Catalyst O2 Cohort at Outcast Ventures. I’m looking for wildly ambitious people to work with on technical AI problems. I’m always up for coffee around SF or Palo Alto, please reach out at [my first 2 letters][my last name] [at] [a university] [dot] [this page's logo].
I’m trying to figure out how to create fully automated companies. I’m currently into how to create a fully automated company. See the list of projects I’m building so far.
Startups
2026
Hippocratic AI — VP Strategy, focused on accelerating AI adoption and automation across the company, and developing a new product called AI Orchestrator.
Explainer Videos — A library of skills for agents to create explainer videos, with Manim animations and screen demonstrations.
AgentFeed — Andrej Karpathy tweeted that explainer videos are the best solution for the wall of text your agents are sending you. So I built a TikTok-like video feed for your agents to communicate with you.
Howard Health — Instinct for health. Your personal health AI agent.
ceo — One conversation to coordinate AI agents across multiple projects. Built on top of CompanyOS.
obsrv — Profiles AI inference servers to help identify performance bottlenecks and guide optimization.
Inferencio.com — This is my attempt to revive Llama API.
CompanyOS — A template for an agentic company, with defined roles and an operating cadence so the company can run itself.
edreis.ai — My personal bio page, the site you’re reading now.
Finegrain — Shared company memory and a model that learns with your team. 4th place at the Gbrain/RiverAI Hackathon.
2026
HealthMCP.com — Connects LLMs to on-demand doctor consultations, lab tests, and prescriptions, with a doctor in the loop.
2025
ChatHealth.com — An iOS app to track all my health data with LLMs. I lost 20 pounds while building it (unfortunately, I gained it all back after I stopped, so maybe I should restart).
2023
Llama API — An API for building with open language models.
MultiOn — Stanford’s pioneering computer-use agents in 2023.
CoScribe.co — An autocomplete tool for EHRs, built using the recently launched ChatGPT.
Research Highlights
2025
Merlin — A vision–language foundation model for three-dimensional CT scans.
MedHELM — A framework for evaluating language models across medical tasks.
MIRIAD — Millions of medical question–answer pairs grounded in scientific literature.
2024
CheXAgent — A foundation model for chest X-ray interpretation.
AgentClinic — Evaluating AI agents in simulated clinical environments.
2023
MedAlign — Clinician-written tasks for evaluating AI on electronic medical records.
MedFlamingo — A multimodal medical model that learns from a few examples.
2022
BRAX — The first open chest X-ray dataset from Brazil.