Postdoctoral scholar at Lawrence Berkeley National Laboratory, working at the intersection of biochemistry, metabolic engineering, and machine learning. Currently building high-throughput bioprocess pipelines for rare earth element biorecovery.
I'm a biochemist and metabolic engineer who builds bridges between the wet lab and computation.
At Lawrence Berkeley National Laboratory, I lead a program on rare earth element biorecovery using Methylobacterium extorquens AM1 and drive high-throughput bioprocess development on Hamilton Vantage liquid handlers. Before LBNL, I earned a PhD in Chemistry from Texas A&M where I characterized metalloenzymes in the lumichrome catabolic pathway and studied C–S bond-forming cytochrome P450s.
I'm also pursuing an M.S. in Computer Science (Machine Learning) at Georgia Tech — deepening my work in predictive modeling, cheminformatics, and graph neural networks for substrate utilization prediction.
Most of what I build lives on GitHub — agents and skills that put real experimental datasets in reach of a natural-language question, plus the analysis pipelines behind my benchwork. A few I am happy to point at:
Gene-fitness data for P. putida, answerable in plain English.
A Python API and Claude Code skill over RB-TnSeq fitness data from the LBNL Fitness Browser — 332 competitive-fitness experiments, 4,732 genes scored. Ask what a locus does and get an answer grounded in the measurements: condition-specific phenotypes, cofitness neighbours, and fModule membership, instead of a guess from the gene name.
View on GitHubA multi-agent panel that red-teams a grant proposal against its call.
A Claude Code skill that runs the review the way a real panel would: deconstruct the call, research the prior art, critique the figures with vision, audit the proposal for internal discrepancies, convene a multi-persona reviewer panel with adversarial QA, then an Area Chair meta-review — ending in a scored, citation-grounded report. Supports both a fresh review and a revision / red-team pass against a previous version.
View on GitHubFrom a t-test spreadsheet to a biological story, in eleven stages.
An end-to-end microbial proteomics skill: differential expression calling, UniProt annotation, GO / KEGG over-representation, pathway maps with up/down overlay, STRING interaction networks with hub and community analysis, AlphaFold structural inspection, PTM site annotation, and a literature-grounded synthesis report — with human approval gates at the decisions that matter.
View on GitHubType a molecule, get a prediction — cheminformatics in the browser.
A prototype web app that takes a chemical name or SMILES string and returns a prediction from a pre-trained graph neural network, wrapping a model that normally needs a notebook in an interface a bench scientist can actually use. Built during a STRUDEL AI workshop and featured in the STRUDEL gallery.
View in the galleryPCR, cloning, Gibson & Golden Gate assembly, CRISPR-Cpf1, CRAGE, gene insertion / deletion in P. putida KT2440.
Recombinant expression & purification, enzyme kinetics, assay development, AlphaFold, PyMOL, AutoDock Vina.
Screening, optimization, and independent X-ray diffraction data collection.
HPLC, LC-MS method development, GC-MS (SRM/MRM), NMR, UV-Vis, FPLC (ÄKTA), plate readers.
Python, Java, supervised / unsupervised ML, AutoGluon, graph neural networks (Chemprop D-MPNN), cheminformatics, JMP.
Hamilton Vantage liquid handling, 96-well plate assay development, Biolog PM phenotype microarrays.
Lawrence Berkeley National Laboratory · Berkeley, CA
Georgia Institute of Technology · Atlanta, GA (online, part-time)
Texas A&M University · College Station, TX
Lahore University of Management Sciences · Pakistan
Manuscript in preparation · Graph neural networks + Biolog PM data
Manuscript in preparation
ACS Central Science, 11, 2353–2365.
Biochemistry, 58(14), 1837–1840.
Current Opinion in Chemical Biology, 47, 134–141.
I read a lot — mostly Asian literature. Over the years I've enjoyed exploring a range of hobbies, from bouldering, softball, and football to, more recently, crocheting.
Especially Asian literature — Hosseini, Murakami, Pamuk.
Problem-solving with your body — hard to beat.
Weekend league, Bay Area.
The round-ball kind. Lifelong habit.
My newest obsession — meditative and satisfying.
The leadership playbook of Silicon Valley's Bill Campbell — practical, human, and surprisingly tender.
Drive, When, To Sell is Human — smart, behavior-science-grounded writing on motivation and work.
On why passion and perseverance matter more than raw talent — a lens I come back to often.
The Kite Runner, A Thousand Splendid Suns, And the Mountains Echoed — deeply moving, unforgettable storytelling.
Live contribution heatmap, pulled in real time from github.com/saadnaseem — updates whenever I push.
Happy to chat about enzyme mechanism, synthetic biology, applied ML in the life sciences, or new collaborations. Email is the fastest way to reach me.
snaseem@lbl.gov