Openings at Genolux
We're a small team in Baltimore building computational tools for antibody engineering. If you're a computational biologist, ML engineer, or full-stack developer who wants to work on problems that matter to drug discovery, read on.
Open positions
You'll work on antibody-antigen docking, CDR loop modeling, and the scoring model calibration pipeline. We need someone with working knowledge of Rosetta and/or OpenMM, and experience interpreting protein-protein interaction energy landscapes. A background in antibody structure or immunology is a strong plus.
Work on the scoring model layer: protein language model integration, transformer architectures for sequence representation, and calibration of the ΔΔG regression pipeline against experimental data. PyTorch required. Experience with ESM-2, AntiBERTy, or similar biological sequence models is directly relevant.
You'll own the product surface: the API, the web interface, job management system, and output delivery layer. We need someone comfortable with scientific data formats (PDB, FASTA, JSON schemas) who can build clean tooling for PhD-level users. TypeScript and Python preferred, scientific SaaS experience a plus.
Deep biologics discovery background and relationships in pharma or biotech. You'll help early customers get maximum value from the platform and identify where the platform needs to expand. This is not a sales role — it's a scientist who can also represent Genolux in technical conversations with antibody discovery teams.
How we work
01
Five people. No management layers. When you find a problem, you fix it. When you disagree with a design choice, you say so. That's the entire process.
02
We release benchmark datasets and evaluation code publicly. If you do work here that's scientifically significant, it goes on bioRxiv with your name on it. We build credibility through transparency, not through secrecy.
03
Our early customers are PhD-level antibody engineers at small biotechs. You'll talk to them directly. Their experimental results against our computational predictions are the feedback loop that improves the model. No ticket queue.