Ali Shahmohammadi Ph.D.
Applied AI Engineering · Scientific Data

I make scientific data trustworthy for AI.

Agentic AI that designs experiments, physics-informed models that learn from them, and FAIR data foundations that make it all reliable — from a decade in pharmaceutical R&D.

Read the writing → View résumé
ML FEEDBACK → AGENTIC designs exp. informs ML Physics-Informed ML PINNs · Neural ODEs Neural Operators DeepONet · FNO ∇²u=f Physics Priors Design of Experiments Sequential DoE Bayesian Opt. Active Learning Efficient Exploration Agentic AI Systems Multi-Agent Pipeline Human Oversight Orchestration · Eval Autonomous · Audit Data Foundations FAIR · Ontologies · Semantic Layer · Master Data · AI Governance
10+
years bridging science & AI
20+
open-source projects
12+
production agentic systems
25+
published articles

What I work on

Three threads, one goal: data AI can rely on

The writing below is organized around these, so you can jump straight to the track you care about.

Physics-Informed ML

Embedding differential equations in neural networks to learn from sparse, noisy data — a full beginner-to-advanced series on PINNs and neural operators.

Scientific ML →

Agentic AI Systems

Multi-agent pipelines with human oversight — ontology curation, first-pass data governance, and audit-ready decision trails in regulated R&D.

Agentic AI →

Trustworthy Data & FAIR

Making R&D data findable, interoperable, and AI-ready — FAIR maturity, semantic layers, master data, and AI governance you can actually operate.

Data Governance & FAIR →

Writing

Articles & projects

Filter by topic, then sort for where you are — a learning path from the basics, or the latest first.

Topic
Sort

Career

A decade turning science into systems

Ali Shahmohammadi

I work at the seam between rigorous science and shipped software: taking the messy, sparse, hard-won data of pharmaceutical R&D and building the pipelines, agents, and evaluation harnesses that let AI use it without making things up. My background in physics-informed modeling keeps me honest about what models can and can't know.

Sep 2022 — Present
Associate Director, Applied AI Engineering & Scientific Data
Takeda Pharmaceutical Inc.

Agentic AI for ontology curation, MCP-based scientific data connectors, enterprise FAIR data strategy, and a digital twin for continuous API manufacturing. Research partnerships with MIT, BYU, Brown, and Purdue on PINNs.

Aug 2021 — Sep 2022
Senior CMC Scientist
Moderna Inc.

ML models for mRNA drug-substance stability and shelf-life prediction; IVT reaction optimization combining ML and mechanistic modeling; contributed to IND and BLA submissions.

Oct 2019 — Aug 2021
Post-Doctoral Research Fellow
The University of Texas at Austin

Fundamental models for thin-film etch optimization; model-based DoE tooling in R Shiny; first-principles viscoelastic modeling; molecular-dynamics simulations at 100k+ particle scale.

2019
Ph.D., Chemical Engineering
Queen's University · Kingston, ON, Canada

Process systems engineering & statistical design of experiments.

Full résumé →

Contact

Let's talk about trustworthy AI for science

Open to conversations on agentic systems, scientific data infrastructure, and physics-informed ML.

Email me LinkedIn GitHub