gadi cohen

I build AI agents that run in production and take real actions. Most of my work is taking them from 0 to 1.

Granted Health, founding team · ex-Assort, Blackbird.AI · Brown CS

Answers come from my real work.

Currently

Granted Healthfounding team · staff engineer · 2025–now

Building the AI agent system that resolves complex medical-billing issues for patients. I own the agent architecture, retrieval, infrastructure, and evaluation.

Multi-agent case automationReal-time context engineProduction RAG over documentsTemporal durable executionEvals & observability
Read the deep-dive →
01

Work

Four years building production AI in healthcare: voice agents, RAG, multi-agent automation. On a decade of shipping full-stack products.

Assort Health

2023–24

Founding engineer on a voice-AI contact-center platform for healthcare. Shipped intent routing that lifted automation from 70% to 85%, and ran infra for thousands of concurrent LLM patient calls.

ElevenLabs · OpenAI · Deepgram · Kubernetes

Blackbird.AI

2022–23

Senior engineer on a disinformation-detection platform. Built an LLM-integration prototype in under a month that helped secure funding. Cut network-viz generation time by more than half.

Python · React · Elasticsearch · AWS · Celery

Sisense

2019–22

Built a Python/R computing platform and architected RBAC plus audit for Periscope Data.

Ruby on Rails · React · Python

pMD

2016–18

Full-stack and Android for a clinical charge-capture app serving thousands; built HL7 integrations.

Java · Android · MySQL · HL7

02

Side projects

DreamDraft

● live

AI children's-book creator. Fine-tunes Stable Diffusion on a family's own photos so the kid is the hero of the story.

Hard part: personalized image generation reliable enough for non-technical parents.

Pixelbench

○ building

Evaluation and monitoring for text-to-image models: measure quality, compare models, catch regressions.

Hard part: turning subjective image quality into numbers a team can act on.

Olaa

○ building

Language-learning journaling app. Intermediate learners practice answering real questions in their target language.

Hard part: feedback that corrects without killing the habit.

03

Contact

Open to talking with people building ambitious AI.

Gadi Cohen