Engineering the Summer is an annual series following UChicago Pritzker Molecular Engineering students as they embark on summer internships and career experiences.
UChicago Pritzker School of Molecular Engineering (UChicago PME) PhD student Alex Berlaga was looking for a way to turn his computational and AI skills to discovering new drugs and therapies.
The third-year student in the Ferguson Lab found his opportunity through a summer internship with Merck, taking his expertise to one of the world’s largest pharmaceutical companies.
He shared some of his experiences for UChicago PME’s yearly Engineering the Summer series:
What first sparked your interest in your area of study?
I have wanted to pursue research relevant to drug discovery since early in college when I found out members of my family had diabetes and various cancers. Since then, I have wanted to apply my best skills (math and algorithmic thinking) toward a career where I could help make innovative medicine.
What research are you focused on at UChicago PME?
I develop and apply physics-guided computational methods that blend machine learning with molecular simulation. My work spans building and refining transferable force fields, engineering synthetic peptidomimetic compounds and collagen-mimetic peptoids for collagen-mimetic character and silica binding, and using Flow Matching framework for generative protein side chain reconstruction. Across these projects, I focus on integrating high-performance computing, active learning, and simulation to accelerate the discovery of biologically relevant macromolecules.
What has been your experience so far this summer at your internship?
My project involves a lot of similar active learning work to my PhD, but applied to genomic datasets for target discovery. I have been learning a lot of new information about omics and sequencing, but also working with a lot of familiar methods.
What do you find most exciting about interning at Merck?
I’m excited by how seamlessly my PhD work maps onto real-world target-discovery pipelines in drug development. For the first time, I can see the acquisition functions and uncertainty-aware models I’ve built guiding concrete decisions to shorten experimental cycles and sharpen therapeutic hypotheses. It’s deeply rewarding to contribute methods that are immediately relevant to industry needs while still preserving the open-ended, exploratory spirit that drew me to research in the first place.
What impact do you think your field will have on the world in the next 10 to 20 years?
I don’t have a strong clue about where Bayesian optimization specifically will be in 10-20 years, but ML in general is clearly taking over every industry including therapeutics. While the use of ML in the early-discovery, preclinical phase is more straightforward, I could see it shortening the drug development pipeline even in the latter parts clinical stage by optimizing drug combinations and clinical trial recruiting to reduce the cost of getting to an equally successful FDA approval.
What role do you hope to play in that vision of the future?
I think my passion is still in applying ML to early drug discovery. New methods for protein-ligand interaction modeling are literally coming about every month, and I think we are close to the point where they move from tech companies, startups, and universities to places with the resources to actually apply them within an actual drug development pipeline. I would love to contribute to that.
What else would you like people reading the article to know about you, your internship or your research area?
Machine learning is opening drug discovery to far more researchers than ever before. Open-source libraries and publicly released protein and molecule language models mean that students and faculty with solid algorithmic skills can now run virtual screens, set up Bayesian-optimization loops, and share data with wet-lab collaborators without massive infrastructure. The work still demands rigor in chemistry, statistics, and model validation, but the entry points are increasingly accessible.