Protein Production Scientist, Biologics

The opportunity Substrate is building the protein production cascade that turns a sequence into characterised, quality-controlled protein at scale, and you will build it from the first manual run. This is the work upstream of every protein data point Substrate generates: DNA design, molecular biology, protein expression, purification, and quality control, developed by hand and engineered to move onto automation. The data this cascade produces is the product. Substrate is the critical infrastructure layer between AI and biology, and biological foundation models can predict but cannot experiment; the high-quality, large-scale data they need does not exist yet. Your work at the bench is what brings it into existence.

About Substrate Substrate is building the critical infrastructure layer between AI and biology: an AI-native automated lab that produces biological data at scale. AI for biology has a data problem, not a compute problem. Biological foundation models can predict but cannot experiment, and the high-quality, large-scale data they need does not exist. Substrate generates it, with quality and provenance built in. We are not a CRO and we are not a cloud lab. The company was founded by four co-founders and is funded through a combination of equity and debt. The first lab is in London, with a larger automation node to follow. The work starts with two scientific verticals, protein characterisation and functional genomics, and this role sits at the centre of the protein work.

The role You will build the protein production cascade that sits upstream of molecular characterisation and functional assays: DNA design, molecular biology, protein expression, purification, and quality control through analytical characterisation. The cascade will eventually run autonomously on Substrate's automation platform. In the first phase, you develop each step by hand, running protocols manually, setting reproducibility and quality thresholds, and proving each phase works before it moves onto instrumentation. As the automation platform comes online, the work shifts toward instrumented execution, equivalence validation, and the engineering judgement calls that decide which manual steps get automated and which stay in human hands. You will work directly with the automation engineering and software teams on the boundary between scientific protocols and autonomous execution, at Scientist or Senior Scientist level, alongside Principal Scientists at the bench, and you will grow into protocol authorship over the first year.

What you will do in your first twelve months Phase 0 - NOW to AUG 2026 Land in the lab. Set up your bench at the London site and start manual assay development alongside a senior member of the protein sciences vertical. Get hands on the first priority protein production workflows as the day-one assay menu locks. Start building DNA design and production (linear DNA), protein expression (cell-free), and protein purification and QC at high throughput. Run them by hand on equipment that will eventually move onto the automation platform, capturing the data-structure and metadata decisions that translate into automation design. Build reproducibility, precision, and acceptance thresholds into the workflow. Contribute to the day-one menu decisions and begin authoring SOPs for your slice of the assay portfolio. Help interview the scientists and lab technicians joining alongside you.

Phase 1 - SEP to DEC 2026 Have a linear DNA, cell free expression, purification, and QC workflow in a semi automated state, running at the scale of hundreds to thousands per week and ready for full automation. Co design protocols with the software and automation engineering teams so the manual versions you validate are automation ready by design. Decide which manual judgement calls have to be engineered out before they reach the platform. Start building DNA design and production (plasmid DNA), protein expression (mammalian), and protein purification and QC at high throughput, again by hand on equipment that will move onto the platform. Build reproducibility, precision, and acceptance thresholds into the workflow. Contribute to co design conversations with the first commercial customers, including the foundation model partners coming online from 2027.

Phase 2 - JAN to MAR 2027 Workcells arrive in the lab. Move the validated assays onto them, running with instrumentation and human intervention in the loop. Validate equivalence against your manual baselines and triage failures. Open the assay menu to customers through manual and semi automated services. Run real experiments for real customers. Help bring on the next scientists and lab technicians as the vertical grows.

Who you are You are a protein scientist who is excited about the actual work: designing, running, and validating at the bench. You are comfortable in the details. You have hands on experience with high throughput, plate based methods across molecular biology, protein expression, purification, and QC. The shape of the problem is what attracts you: assays designed for autonomous execution from day one, in a business where the data the lab produces is itself the product. You write good SOPs, and you hold yourself and your colleagues to clear reproducibility thresholds. You are pragmatic about being hands on in the early phase, when the cadence is heaviest, and you understand it eases as protocols move onto instrumentation. You enjoy working at the boundary with non biologist colleagues (automation engineers, software engineers, AI researchers), and you do not require them to be scientifically fluent before you will collaborate. This role is open at Scientist and Senior Scientist level. The work is hands on bench science at both, with collaboration into automation and software; the difference is depth of ownership, design authority, and how much of the team you help build. We do not hire people into boxes, and the early team stretches beyond the strict edges of any role.

Must Have Comfortable handling DNA design and linear and plasmid DNA production at high throughput, expressing proteins, and purifying them with QC steps for validation, executing protocols at the bench through manual, semi automated, and instrumented phases. Core experience within biologics. A track record of working alongside non scientist colleagues (automation, software, computational) on a shared workflow. A PhD in protein biochemistry, molecular or structural biology, biophysics, analytical characterisation, or biologics developability, with two or more years of relevant hands on experience; or a relevant science MSc with five or more years of equivalent bench experience in protein science. Independent, hands on competence across protein science: molecular biology (DNA production with Gibson and Golden Gate assembly at transfection grade quality), protein expression (cell free, IVTT, E. coli and mammalian), and protein purification (automated magnetic bead workflows), with exposure to at least one of these at high throughput (24, 96, and 384 well plate formats). Protein characterisation across at least two core biophysical methods (for example CE SDS, DLS, nanoDSF/DSF, analytical SEC, HIC, or LC MS intact mass analysis). Confident data analysis, including telling instrument artefacts apart from genuine molecule behaviour. Fluency with structured experimental data capture, and proficiency with a LIMS, ELN, or analogous data infrastructure. Ready to grow into protocol authorship and SOP ownership over the first twelve months.

Nice to Have Direct experience moving biophysical or analytical assays from manual workflows onto lab automation platforms (at least a handoff into an automated platform). High throughput experience in at least one core biophysical platform (SPR or BLI for binding kinetics), plus one or more orthogonal characterisation or developability assays (for example nanoDSF/DSF, DLS, analytical SEC, HIC, AC SINS, cIEF, or CE SDS). Experience developing novel high throughput technologies in DNA generation, cell expression (cell free, mammalian), or bead based purification. Method development and validation experience: defining acceptance criteria, references, and controls that other scientists have run successfully. SOP and protocol authorship that others have executed, and experience supervising at least one junior scientist or technician. Experience working with computational or AI/ML colleagues on closed loop assay programmes. Background at an AI native biotech or foundation model company.

Why this is unusual Most protein science roles like this inherit an established workflow and optimise it. This one does not. No assay in this vertical is being retrofitted onto automation; every protocol is designed for AI in the loop execution from the first manual run. The data each assay produces, its output structure, metadata, provenance, and consistency across runs, is treated as a first class scientific constraint, because that data feeds directly into foundation model training pipelines. Your decisions at the bench affect what the orchestrator has to do and what data leaves the building. Some scientists find this energising; others find it outside the lane they trained for. It is worth knowing in advance which one you are.

How we work This is an in person, lab based role at 20 Triton Street in London. The manual development phase is hands on.