Pharma R&D
AI Engineering at Roche × Chugai
Building AI pipelines for scientific use in the cloud. Bridging the gap between scientific understanding and AI engineering.
Muhammad Mamduh Ahmad Zabidi, PhD
I work at Roche on AI systems for pharmaceutical R&D, solving problems where biology and engineering meet.
My background covers biochemistry & molecular biology, computational biology, and AI engineering. I did my PhD in computational biology at the Institute of Molecular Pathology in Vienna, Austria, and bachelor's degree at Purdue University, USA.
About
I started my career at the bench in cancer research, studying genes that are involved in nasopharyngeal cancers. I then did a PhD in computational biology at the Institute of Molecular Pathology in Vienna, working on gene regulation and genomics.
Over time I moved toward the engineering side, working on data infrastructure and AI systems.
Today I work at Roche in collaboration with Chugai, building and evaluating AI systems for scientific R&D. I spend my time answering scientific questions through engineering.
Selected work
Pharma R&D
Building AI pipelines for scientific use in the cloud. Bridging the gap between scientific understanding and AI engineering.
Scientific AI Systems
Built a system to convert regulatory documents into structured knowledge using multimodal LLMs, enabling downstream machine learning workflows for therapeutic design.
Edge AI Engineering
Built an offline desktop AI system for video intelligence, combining OpenVINO-optimized models, Tauri, gRPC, SQLite memory, agentic routing, and local report generation.
View projectGenomic AI
Built a biologically grounded framework for evaluating genomic foundation models, and benchmarking against experimental evidence.
Publications & patents
First-author publications in computational biology, published in Nature, Nature Biotechnology, and Trends in Genetics.
Co-led a large-scale Asian breast cancer genomics study, published in Nature Communications.
Co-inventor on patents in cancer detection and model routing for AI systems.
Full publication record on ORCID and Google Scholar.
Selected writing
Regulatory AI
An essay on FDA approval documents as a public, semi-structured archive of biomedical evidence, regulatory judgment, and evolving standards in modern medicine.
Read essayScientific Data Engineering
An overview of how scientific and clinical data can be sourced, including in-house generation, partnerships, vendors, and public datasets.
Read essay