4 WEEKS | LIVE (ONLINE) | ADVANCED LEVEL | STARTS: FEB 7, 2026
You’ll build real molecular generative models, train them, evaluate them, and deploy a working design pipeline. The curriculum is designed for professionals working in drug discovery and crop sciences who want practical, deployable skills in molecular generative AI. PhDs and postdocs are welcome, but the bar is set for industry level workflows and expectations.
HOSTED BY
Pankaj Mishra, PhD - I’m the Co-founder and CTO of Future Therapeutics, an AI-first biotech based in Berlin, where I build proprietary AI systems for drug discovery. I hold a PhD from the University of Freiburg and have worked in small molecule AI since before it became mainstream, back in 2018 I was already building deep learning algorithms for “ultra-large chemical space” when most teams were still relying on classical virtual screening. Over the past few years, my work has been centered on generative molecular design. I’m also a Co-founder of Neovarsity and have taught scientists and engineers across biopharma, including teams from J&J, Bayer, Takeda, and Novartis. Due to my workload, I now run this program only once a year out of passion. I keep the cohort highly focused and hands on, and I’d love to have you join.
WHAT IT’LL COVER
This is a hands-on, highly technical curriculum, focused on what really matters: data quality, model failure modes, conditioning, and evaluation under real constraints. Below is the week by week structure. Please see the syllabus for week by week structure.
WHO THIS IS FOR
This course is for you if you are a computational chemist or medicinal chemist moving into generative AI, an ML engineer working in drug discovery, biotech, or cheminformatics, a PhD student or postdoc building molecular modeling skills, or a founder or technical lead who wants to build internal capability instead of outsourcing core scientific intelligence. This course is not for you if you are new to machine learning, if you are looking for a no code tool walkthrough, or if you expect “press button, get drug” outputs.
WHAT YOU WILL BE ABLE TO DO
By the end of the program, you will be able to build chemically valid datasets for generative modeling, train and debug multiple classes of molecular generative models, and condition molecule generation on target objectives and real world constraints. You will also be able to evaluate generated molecules using metrics that actually matter in drug discovery, and deliver a capstone level molecular design proposal with defensible scientific reasoning.
PREREQUISITES
This is a highly technical program. You should already be comfortable with Python, basic ML workflows, and cheminformatics, mainly how to handle chemical data. If you need help getting ready, reach out, we offer a prep program.
TOOLS AND STACK
You will work with RDKit for chemistry workflows, PyTorch for model training, and modern molecular generation libraries with practical implementation patterns. You will also build and use evaluation tooling and metrics frameworks similar to what real modeling teams rely on in industry.
