About
How I got here
I am a scientific leader working at the intersection of life sciences and AI. Eleven years at Merck: I started as a chemical engineer with no laboratory experience, spent most of a decade on small-molecule process development, and contributed to eight approved medicines. Today 13 machine learning scientists, data engineers, and chemical engineers report to me, building the models, agent systems, and data products that biologics development runs on.
I write my own algorithms. Almost everything I know about machine learning I taught myself, first through a Georgia Tech masters and a lot of nights. What I have learned since is that adoption is the hard part, not the algorithm, so most of my time now goes to the unglamorous half: the connectors that make lab data readable, and the evaluation that catches an answer which only looks right. All of it is pointed at the same thing it was when I started, which is getting medicines to people faster.
Download the resume
A decade of changing technical roles
Click a chapter for the fuller story.
2026 - Director, Process Modeling & Analytics DSCS Digital
- Leading a 13-person team of machine learning scientists, deep learning specialists, data engineers, and chemical engineers. Five are experienced PyTorch developers; three build agentic harnesses.
- Shipping LLM agent systems now in production for hundreds of scientists, including a multi-agent summarization system whose adversarial grading loop I wrote myself: it scores every draft against ten criteria and makes the generator try again until it passes.
- Directing the data connectors and structured data products that make messy lab notebook, LIMS, and assay data usable by models and applications.
- Founded and sponsor a developers community of practice covering CI/CD, coding tooling, and engineering training across a 150-scientist department.
- Supervising five chromatography mechanistic models now cutting wet-lab resource needs by up to 30% and widening operating ranges by up to 50%.
- Writing the strategy for model-informed decisions that support regulatory filings, including model verification and validation protocols.
2025 - 2026 Director, Data-Rich Experimentation PR&D
- Wrote GARNET end to end in PyTorch and torchdiffeq: a neural-ODE architecture for cell culture and fermentation that describes a process from as few as 30 experiments and holds cross-validated R2 between 0.90 and 0.97 on real programs.
- Applied GARNET to a separate pipeline program, improving titer 15% and harvest-day viability 20% through model-guided feed design; the approach also reduced process-characterization resources by up to 50%.
- Led a 13-person cross-modality team spanning small-molecule catalysis, biologics high-throughput experimentation, and modeling-algorithm development.
- Grew Bayesian optimization adoption to 160 active users and roughly 50 optimizations a year, across more than 20 program steps.
- Deployed a hybrid-model digital shadow at a biologics manufacturing site, supporting deviation analysis, troubleshooting, and real-time plant decisions.
- Received the 2025 ACS GCI Pharmaceutical Roundtable Data Science & Modeling for Green Chemistry award for applying APO to make laboratory experimentation and manufacturing processes more resource-efficient.
2022 - 2025 Associate Principal → Principal Scientist PR&D Data-Rich Experimentation
- Founded Algorithmic Process Optimization, bringing closed-loop Bayesian optimization into Merck's pipeline, and published obsidian, the department's first open-source codebase.
- Onboarded 100+ scientists across small-molecule and biologics programs, cutting wet-lab resource use 25 to 75% per campaign.
- Improved an mRNA in-vitro transcription process more than 100% across 12 variables in 43 experiments.
- Co-developed GARNET, a neural-ODE model line that moved from enzyme kinetics into upstream bioreactor modeling.
- Received a 2024 ACS Division of Organic Chemistry Technical Achievement award.
2019 - 2022 Senior Scientist, Reaction Engineering CERD
- Redesigned the molnupiravir API reaction in four weeks under pandemic pressure, using high-throughput experiments, process models, and simulation to support a new control strategy with a more than 10x stability window and an order-of-magnitude lower impurity burden.
- Represented Merck as its first lead engineer for synthetic electrochemistry, then delivered the industry's first kilogram-scale example, later named OPRD's Publication of the Year.
- Built nitrosamine risk-assessment tools for 12 commercial products, saving more than $1.5M in outside testing.
- Self-taught data science and Python while starting a master's degree part time, then began teaching the department.
2015 - 2019 Associate Scientist → Scientist Chemical Engineering R&D
- Technical lead across synthetic steps and separations for doravirine, now approved as Delstrigo and Pifeltro.
- Delivered 600+ kg of clinical-supply API during a pre-validation campaign in Ballydine, Ireland, and authored 150 pages of the NDA process reports.
- Won back-to-back CERD Technical Achievement Awards in 2015 and 2016.
Awards
- ACS GCI Pharmaceutical Roundtable, Data Science & Modeling 2025
- ACS Division of Organic Chemistry, Technical Achievements 2024
- OPRD Outstanding Publication of the Year 2022
- ACS & EPA Green Chemistry Awards (doravirine, gefapixant, molnupiravir) 2018-22
- Merck MS&T Innovation Award, runner-up 2021
- Merck CERD Technical Achievement Award 2015-16
Invited talks
- Keynote, Industrial Applications in Computing & Systems Technology (CAST) 2026
- GARNET bioreactor digital twin, Recovery Series Modeling Workshop 2025
- Algorithmic Process Optimization, ACS GC&E and Kinetic Modeling Roundtable 2025
- Mechanism & kinetics in the development pipeline, Mettler Toledo webinar 2023
- Belzutifan transfer hydrogenation, Mettler Toledo webinar 2022
Service
- Founder and sponsor, Developers Community of Practice 2026
- Programming Chair, AIChE PD2M Forum 2026
- Session Chair, AIChE Student Conference on Pharma Chemical Engineering 2026
- Track Chair, Predictive Scale-Up/Scale-Down, AIChE PD2M Forum 2024-25
- Scientific Advisory Board, Technobis ReactALL 2024
- Steering Committee, Mettler Toledo Kinetic Modeling Roundtable 2021-25
Away from work, I spend nearly all of my time with my wife and two daughters. We play in the pool, go to the beach, and visit family. I've played piano my whole life, and have been playing more classical piano as an adult. Liszt has taken up a lot of that time. I also listen to science, technology, and history podcasts, and a lot of comedy. I also keep building things for fun, which lately has meant game development in Unity and Blender and writing agent harnesses on my own time.
(Speaking of piano: there's a short one hiding on the home page. Click my name.)
A few examples from the work.
Read the case studies