I'm a bio- and software engineer working at Imperial College on DNA design. In uni, I discovered that you can use computers to make biodesign more predictable. Since then, I cold-called my way into a wet-lab to learn DNA assembly and am teaching myself machine learning to build the computational methods for the next generation of synthetic biology.
Projects
tropical
Autoregressive mRNA language model conditioned on protein sequence and translation efficiency
yorzoi
Predict RNA-seq coverage from DNA sequence
plextract
Extract line chart data from plots
breseq-on-modal
Read-alignment with breseq in the cloud
List of {R,D}NA sequence models
A recurringly updated list of machine learning models trained on RNA or DNA sequences
Publications
Writing
Resume
Education
Working on generative models for expression-conditioned DNA sequence design. Supervised an M.Eng. student developing machine-learning models to predict growth rates (GitHub).
Key Coursework: Theoretical Systems Biology (ODEs & Michaelis-Menten kinetics), Opentrons Automation Workshop.
Selected for CDTM (≈ 5 % acceptance rate), an entrepreneurship program whose alumni have founded nine unicorns.
Key Coursework: Introduction to Machine Learning, Linear Algebra, Discrete Probability Theory.
Work Experience
Worked with the CTO to build the initial MVP (Chat Interface and RAG pipeline) using React, Next.js, tRPC, and GraphQL; rolled out to three pilot customers.
Cloned 30 plasmids to help build an ATP-auxotrophic E. coli strain using PCR, Gibson/Golden Gate/LCR-Cloning, and other molecular biology techniques.
Built a domain-specific language for protein design including DSL specification, compiler, web interface and deployment using Python, React.js, ANTLR, Docker, and GCP (Cloud Run/Storage/Build, IAP), Terraform.
Evaluated ML models for their explanatory power of rare variant effects on aberrant gene expression using Python, Bash, and Snakemake.
Led a four-person, €20 K consulting engagement for an electronics supplier, delivering a market research study.