H.i.
Welcome!
Founded by Nate Harms, PhD in March, 2024 as a blog…
Harms Informatics (H.i.) is an independently-owned scientific consulting firm operating at the intersection of computational chemistry, data engineering, and applications in AI. We work with our clients to identify their unique chemistry needs and deliver custom-made AI solutions using modern software developing principles.
Our mission is to accelerate breakthroughs in chemistry by building robust AI solutions.
How will H.i.
accomplish this?
Through independent research and collaborative partnerships, H.i. hopes to use our skills to help democratize machine learning (ML) for chemical property prediction.
By sharing data, models, and best practices with the broader scientific community, we will enable others to learn more, move faster, and discovery incredible things.
What are our specialities?
-
Developing chemical data pipelines is our bread-and-butter. When designing a pipeline, Nate is an expert in gathe requirements, designing solutions, and delivering repeatable and reusable data pipelines for chemistry.
Related projects include…
Enko’s DEL enumeration pipeline: Translates DNA barcodes into chemical building blocks, applies reaction schemes to building blocks, in order, to generate fully enumerated DEL instances.
Valo Health’s GOSTAR curation pipeline: Queries Valo’s version of the GOSTAR small molecule database for relevant assay data describing protein inhibition and ADME assays. Corrects, harmonizes, and refines assay data using configurable string- and substructure-filters to generate curated datasets and labeled training sets.
-
We build analysis workflows that turn complex chemical data into clear, interpretable insights. Our work spans assay harmonization, molecular property analysis, SAR exploration, and the creation of visualizations that help chemists and data scientists quickly understand what the data imply. Everything we deliver is reproducible, extensible, and designed to integrate seamlessly into existing discovery processes.
Representative capabilities include…
Interactive plots for SAR trends, chemical embeddings, and property distributions
Automated exploration of outliers, descriptors, and chemical subspaces
Data-cleaning and feature-engineering pipelines that feed into modelling workflows
-
Machine learning creates value in drug discovery only when models are reproducible, interpretable, and operationally reliable. Harms Informatics helps teams bridge the gap between exploratory modeling and production-ready AI by designing MLOps workflows tailored to chemical and biological data. Our focus is on systems that integrate cleanly with discovery workflows, use chemically meaningful evaluation metrics, and remain trustworthy as data and projects evolve.
Representative capabilities include…
End-to-end model lifecycle management (data → training → validation → deployment)
Versioned datasets, features, and reproducible training pipelines
Model evaluation grounded in chemically relevant performance metrics
Monitoring for data drift, domain shift, and model degradation
Integration of ML systems into existing discovery and decision-making processes
-
Clear communication turns strong science into confident decisions. Harms Informatics helps teams distill complex chemical data, analyses, and models into narratives that are accurate, concise, and audience-appropriate — without sacrificing scientific rigor.
Representative capabilities include:
Scientific slide decks for project reviews, strategy discussions, and external presentations
Clear data visualizations that surface key insights without oversimplification
Technical storytelling that connects data, methods, and conclusions into a coherent message
Audience-aware communication for technical, cross-functional, and non-technical stakeholders
About the owner,
Nate Harms
Nate Harms comes with 10+ years of experience in scientific software engineering for processing and modeling chemical data. Nate obtained his BS in Chemical Engineering at Oregon State University and his PhD in Chemical Engineering at Northeastern University.
Nate began coding during his junior year of undergrad during an internship with HP Inc. At HP, he used learned Python and created a rudimentary database of inks. Although Nate’s undergraduate thesis was primarily wet lab research, Nate’s experience at HP motivated him to refine his coding skills in grad school. Under Professor Richard West, he became skilled at Python and developed software to perform automated transition state theory calculations for elementary reactions common in combustion. He has attended workshops on best practices for software carpentry and data science, and presented first-hand research at AIChE, ACS, and many other regional conferences.
Following graduate school, Nate entered the biotech industry at a startup called ZebiAI. Since then, he has worked at multiple biotech companies to construct chemical data pipelines for tasks ranging from data curation, to library enumeration, to ML modeling. Professionally, he considers himself something like an “Applied AI Engineer”, but ultimately, he is a dynamic scientist who uses machine learning to execute on cheminformatic projects.
Over the years, Nate has acquired deep knowledge on many domains of chemistry. Namely…
Combustion modeling
Reaction mechanism generation and pathway analysis; QM geometry optimization; thermodynamic and kinetic property calculation
DNA-encoded library informatics
Hit identification; in vivo confirmation; Physical-Chemical property optimization; cross-species safety and toxicity consideration
Agrochemical Discovery
Hit identification; in vivo confirmation; Physical-Chemical property optimization; cross-species safety and toxicity consideration
Small Molecule
Drug Discovery
Hit identification; hit to lead & lead optimization; ADME profiling; MLOps and Performance tracking
Although Nate has expertise in multiple domains of chemistry, he always enjoys learning something new! He’s confident when he knows something, but he’s also quick to admit if something is confusing. His natural curiosity pushes him to ask questions so he can throughly understand why something is the way it is. He also loves collaboration and is a team player — at the end of the day, he wants to help others and to make an impact.
If you have a project, and you think Nate would be a good fit for it, reach out!