B
Founding engineer
Accepting applicationsBrylo · San Francisco Bay Area
Full-Time Entry AIaiaterf
Posted
2d ago
Category
Test
Experience
Entry
Country
United States
About Brylo:
Brylo is a research-heavy gene editing company. We automate the computational pipeline behind gene editing and build the models that power it — then use the proprietary data that creates to design and license our own gene therapies.
The mission is enormous: design cures and enhance human potential at a speed no incumbent can touch. A goal this ambitious only happens if you're completely all in — so we grind, 24/7, because that's the only way it gets built. We're two founders today, and you'd be one of the first people we bring on.
The Role:
You'd be a founding engineer — which, at this stage, means you build the core system and you help build the company.
The technical core: You'll build the AI/ML systems that make up Brylo's computational pipeline — spanning the entire gene-editing workflow: target discovery, edit strategy, delivery-vector design, manufacturing and cell-engineering analytics, and product QC & safety. Each module is a real ML problem — learning from large, noisy genomic and sequencing data to predict what an edit actually does inside a cell, accurately enough to build a medicine on. You'll work across the whole stack: the models, the data that trains them, and the systems that turn a prediction into a result a scientist can trust.
But we're two people. So you'll also co-author the papers that establish our method and our credibility, work directly with our academic and clinical partners, talk to the scientists who'll use what we build, and help decide what we build next. If you want a narrow, well-defined engineering box, this isn't it. If you want to own a hard problem end to end and have it actually matter, it is.
What You'll Work On:
Build the AI/ML modules themselves. The near-term modeling problems live in safety and QC — variant-aware off-target detection, chromosomal translocation analysis, ancestry-aware variant analysis — real ML systems over messy genomic and sequencing data, held to clinical-grade rigor and built to run at scale.
Bring modern bio foundation models into production. AlphaFold 3, Evo 2, ESM-3, and whatever ships next — wired into workflows scientists actually depend on, not demos.
Design the benchmarks the field doesn't have yet. Read the literature deeply, work out what "good" even means for edited-cell safety and editing outcomes, and build rigorous, defensible benchmarks to measure it — along with the evaluation harnesses, provenance, and reproducibility needed to trust an AI-generated result. This is how we prove the work is real: to ourselves, to partners, and to regulators.
Write the papers. Both the benchmark papers and the methodology papers behind our methods — co-authored with our labs and advisors, aimed at top venues (Nature Methods / Genome Biology tier). Reading, deeply understanding, and writing research is core to this job, not a side task.
Build the proprietary datasets that become the moat. Curate and engineer the data that makes our models better than anyone else's — from reference and truth-set data (e.g., GIAB) to the design-to-outcome data generated as the platform gets used. Whoever owns the data wins this. You'll help own it.
Whatever the company needs next. We're early enough that the single most important problem changes month to month.
Who You Are:
We care about how good you are and how fast you move — not your résumé, your age, or how long you've been doing this. Some of the best people we know are 19. Some never finished a degree. If that's you, good.
Has been exceptional at hard technical things for as long as you can remember — and has the work to prove it. Published research, olympiad medals, top of a serious program, or systems you built that real people use.
Has real depth in ML, and ideally has touched some of: LLMs and agents, computational biology or genomics, multimodal or sequence models, retrieval, or large-scale data systems.
Ships. You'd rather have a working thing today than a perfect plan next month.
Is AI-native. You build with coding agents constantly and you're fast because of it.
Can read a biology paper and a systems paper in the same afternoon and be dangerous in both by evening.
And in temperament — you've been told you're "too intense." You get restless when the problem is easy. You want the thing you work on to be the most important thing you've ever done.
How We Work:
In person, in San Francisco. Same room, every day. You'd be among the first hires. Enormous ownership, no process to hide behind. AI-native — we operate at the frontier of how fast coding agents let a small team move. Due to how ambitious our goal is, we are working on this every waking moment we have.
Compensation:
$150k - 200k base pay
0.5% - 1.5% founding equity
Full-time, in person, in San Francisco.
Show more Show less
Brylo is a research-heavy gene editing company. We automate the computational pipeline behind gene editing and build the models that power it — then use the proprietary data that creates to design and license our own gene therapies.
The mission is enormous: design cures and enhance human potential at a speed no incumbent can touch. A goal this ambitious only happens if you're completely all in — so we grind, 24/7, because that's the only way it gets built. We're two founders today, and you'd be one of the first people we bring on.
The Role:
You'd be a founding engineer — which, at this stage, means you build the core system and you help build the company.
The technical core: You'll build the AI/ML systems that make up Brylo's computational pipeline — spanning the entire gene-editing workflow: target discovery, edit strategy, delivery-vector design, manufacturing and cell-engineering analytics, and product QC & safety. Each module is a real ML problem — learning from large, noisy genomic and sequencing data to predict what an edit actually does inside a cell, accurately enough to build a medicine on. You'll work across the whole stack: the models, the data that trains them, and the systems that turn a prediction into a result a scientist can trust.
But we're two people. So you'll also co-author the papers that establish our method and our credibility, work directly with our academic and clinical partners, talk to the scientists who'll use what we build, and help decide what we build next. If you want a narrow, well-defined engineering box, this isn't it. If you want to own a hard problem end to end and have it actually matter, it is.
What You'll Work On:
Build the AI/ML modules themselves. The near-term modeling problems live in safety and QC — variant-aware off-target detection, chromosomal translocation analysis, ancestry-aware variant analysis — real ML systems over messy genomic and sequencing data, held to clinical-grade rigor and built to run at scale.
Bring modern bio foundation models into production. AlphaFold 3, Evo 2, ESM-3, and whatever ships next — wired into workflows scientists actually depend on, not demos.
Design the benchmarks the field doesn't have yet. Read the literature deeply, work out what "good" even means for edited-cell safety and editing outcomes, and build rigorous, defensible benchmarks to measure it — along with the evaluation harnesses, provenance, and reproducibility needed to trust an AI-generated result. This is how we prove the work is real: to ourselves, to partners, and to regulators.
Write the papers. Both the benchmark papers and the methodology papers behind our methods — co-authored with our labs and advisors, aimed at top venues (Nature Methods / Genome Biology tier). Reading, deeply understanding, and writing research is core to this job, not a side task.
Build the proprietary datasets that become the moat. Curate and engineer the data that makes our models better than anyone else's — from reference and truth-set data (e.g., GIAB) to the design-to-outcome data generated as the platform gets used. Whoever owns the data wins this. You'll help own it.
Whatever the company needs next. We're early enough that the single most important problem changes month to month.
Who You Are:
We care about how good you are and how fast you move — not your résumé, your age, or how long you've been doing this. Some of the best people we know are 19. Some never finished a degree. If that's you, good.
Has been exceptional at hard technical things for as long as you can remember — and has the work to prove it. Published research, olympiad medals, top of a serious program, or systems you built that real people use.
Has real depth in ML, and ideally has touched some of: LLMs and agents, computational biology or genomics, multimodal or sequence models, retrieval, or large-scale data systems.
Ships. You'd rather have a working thing today than a perfect plan next month.
Is AI-native. You build with coding agents constantly and you're fast because of it.
Can read a biology paper and a systems paper in the same afternoon and be dangerous in both by evening.
And in temperament — you've been told you're "too intense." You get restless when the problem is easy. You want the thing you work on to be the most important thing you've ever done.
How We Work:
In person, in San Francisco. Same room, every day. You'd be among the first hires. Enormous ownership, no process to hide behind. AI-native — we operate at the frontier of how fast coding agents let a small team move. Due to how ambitious our goal is, we are working on this every waking moment we have.
Compensation:
$150k - 200k base pay
0.5% - 1.5% founding equity
Full-time, in person, in San Francisco.
Show more Show less