Build & shipprimary
I design your architecture and ship production modules: Phase 1 (architecture plus your first module), Phase 2 (a modular build, one module shipped per month), or an ongoing Fractional AI Partner engagement.
Healthtech AI Engineer
I build healthcare AI systems that ship.
Pharmacist by training, neuroscientist by PhD, engineer by daily practice. I design and ship production AI for biotech, medtech, and digital health teams.
When you want a second opinion first, I also audit, with the same engineering depth. Peer-reviewed in Scientific Reports 2025, with a public demo you can try now.
Proof, not promises
See it, don't take my word for it.
A public app you can try, a peer-reviewed ML paper, and a talk.
Live app
Clinical Protocol Analyzer
Reads a protocol, maps it to ICH-GCP, flags the gaps with the clause cited. Live, paste your own excerpt.
Try the analyzerPeer-reviewedScientific Reports 2025
First-author paper: Self-supervised learning reduces label noise in sharp wave ripple classification.
Read the paperTalkDrug repurposing with AI
IAAM meetup, Marseille, July 2026. From knowledge graph to explainable prediction, demonstrated live. Slides and notebook to download.
Slides & notebookWhat I do
From build to audit, whatever de-risks your project fastest.
Build is the default. Audit and due diligence are entry points when a second opinion comes first.
Audit
Not ready to build yet? I audit your POC, vendor proposal, data readiness, or full AI stack with an engineer's depth. You get architecture diagrams, sample production-grade code, and concrete specs you can act on.
AI Due Diligence
An independent technical read on an AI asset or your own stack, for a fundraise or an investment decision. I surface the real risks, the shortcuts taken, and what it takes to reach production, in terms an investor can act on.
Who I work with
Biotech, medtech, and digital health teams from seed to Series A, with a concrete AI project to build, ship, or de-risk in the next 0 to 6 months. Founders, CSOs, CTOs, CMOs, Heads of Data or R&D who can actually sign. Early-stage pharma R&D welcome.
Why me
Deep science meets shipped systems.
I am Saber Graf, founder of SG AI Solutions. Pharmacist by training, neuroscientist by PhD, engineer by daily practice. I have spent over a decade at the intersection of complex biological data, software engineering, and machine learning, close enough to the science to judge a model, close enough to the code to ship one. I design and maintain an internal cloud-native platform (Django, Next.js, Azure) every day.
Peer-reviewed
First-author paper in Scientific Reports (2025): Self-supervised learning reduces label noise in sharp wave ripple classification. DOI 10.1038/s41598-025-90380-x.
Healthtech-native
Real biological data at scale, with the regulatory and data-governance constraints of biotech, medtech, and digital health.
FAQ
Frequently asked questions
Do you build, or only audit?
I build first. I design your architecture and ship production modules: Implementation Phase 1 (architecture plus your first module), Phase 2 (a modular monthly build), or an ongoing Fractional AI Partner engagement. Audit is one way to start when you want a second opinion before you commit, with the same engineering depth.
What do I actually get?
You get architecture diagrams, sample production-grade code on the critical brick, and concrete specs on data governance, security, and scalability for your exact case, so you finish knowing this person could build it.
Who do you work with?
Biotech, medtech, and digital health teams from seed to Series A, with a concrete AI project to build, ship, or de-risk in the next 0 to 6 months. Founders, CSOs, CTOs, CMOs, Heads of Data or R&D who can actually sign. Early-stage pharma R&D welcome.
What don't you do?
I do not write legal AI Act reports (refer to a lawyer for those) and I do not take CRO engagements, due to a contractual constraint. The technical side of the AI Act (logging, monitoring, explainability) is in scope. Everything else in healthtech AI, ask me.
How do you handle confidentiality and sensitive data?
Every engagement starts with an NDA. No client data is used to train models, and I can work entirely within your infrastructure when required. A background in pharma and healthcare means I understand the stakes around patient data, IP, and regulatory compliance.
Have an AI project to build, ship, or de-risk?
saber.graf@sg-ai-solutions.com
No legal AI Act reports (refer to a lawyer for those), and no CRO engagements (contractual constraint). Everything else in healthtech AI, ask me.