I'm an Associate Principal Scientist in Biologics Process R&D at Merck & Co., working on downstream process development for biologics: chromatography, ultrafiltration/diafiltration, and the molecular and biophysical modeling that supports both. I work at the boundary of wet-lab process development and computational tools, with a particular interest in bringing mechanistic modeling and ML into routine CMC workflows.
🔗 Portfolio: max3925vats.github.io · Google Scholar: profile · LinkedIn: /in/vatsm
| Work | Academic | Fun | |
|---|---|---|---|
| Focus | Downstream process development: chromatography, UF/DF, tech transfer | Molecular dynamics, biophysical modeling, ML for chromatography retention | Tinkering with small computational projects and home lab setups |
| Tools | Python · Hydra · scikit-learn · PyTorch · ChromX/GoSilico-style mechanistic modeling | GROMACS · AMBER · enhanced sampling (metadynamics, REUS) · QSAR pipelines | Ollama on a laptop · a few Raspberry Pis on a home network |
| Modalities | mAbs · fusion proteins · ADCs | Multimodal chromatography ligands · protein-surface interactions | n/a |
| Currently learning | Neural-network featurization of protein structures for PD problems | Bayesian methods for small-sample process data | Edge inference, local LLMs |
I'm a process development scientist who reads code more than he writes it, but I'm working at that boundary deliberately. Most of what lives here is scratch work in Python: data processing for chromatography experiments, modeling pipelines, and small utilities. The polished writeups live on the portfolio site.
If you're a biopharma or computational-bio person interested in how molecular modeling and ML can show up inside CMC workflows, get in touch. Happy to compare notes.