AI transformed digital industries. Biology has not kept pace. The bottleneck is data. The measurements that would train better models sit inside the companies that generated them. Public data is sparse and noisy.

Harell Data puts that data to work. Models come to the data and train there. The data never leaves. Researchers get proprietary antibody-antigen and TCR-pMHC binding data, managed GPU compute, and orchestration in one place. Data providers earn every time compute runs against their data. Model builders keep their models and IP, and can list them on our marketplace.

About The Role

You will be our first Senior Field Application Scientist. You are the scientific advisor to every team training models on Harell. You know the data. You know the models. You help researchers get from first login to a model that works.

This role fits a bioinformatician, computational biologist, or data scientist working on biological data. You want a change in your day-to-day. You want to spend it with customers. You do not want to leave the science.

Why This Role

Stay close to the biology. Work with binding data few people outside the companies that generated it have seen.

Trade one program for many. See the modeling problems of teams across biotech, pharma, and academia.

Become the voice on the data. Alongside our founder, you become the person customers trust to explain what our datasets can and cannot do.

Shape how the field measures progress. Help build blind benchmarks and competitions that score every model to one standard.

What You Will Do

Advise on the data. Explain how our datasets were generated, what they measure, where they are strong, and where they run out. Help researchers match the right data to their model.
Get researchers to their first trained model. Write getting-started guides, sample notebooks, and self-serve resources. Shrink the time from first login to first training job.
Diagnose why a model underperforms. Find the cause, whether it sits in the training code, the GPU setup, or the data. Debug PyTorch and Hugging Face jobs. Tune workloads on H100 and H200 systems.
Bring the science back to the product. Turn customer friction into clear input for engineering. Shape the roadmap.

What You Bring

• Hands-on work in bioinformatics, computational biology, or ML on biological data. Industry or academic. Graduate research counts.
• You have trained and debugged your own models in PyTorch, Hugging Face, or a similar framework.
• You have run training jobs on GPUs or a cluster (SLURM, cloud GPUs) and can explain why a job runs slow.
• You want to work with customers. You get energy from helping other scientists succeed with their models, not only your own.
• You collaborate across teams and groups. You have a track record of working with people outside your own team: other labs, partner companies, core facility users, or cross-functional groups. You are the person others seek out to make hard things clear.
• Clear technical writing. Guides, notebooks, protocols, or docs other people use.

Bonus Points

• Protein, antibody, TCR, immune repertoire, or binding affinity data.
• Moving training code across GPU generations, CUDA and driver versions, or mixed precision.
• AWS or GCP, Linux, Docker.
• Startup or zero-to-one experience.
• Consulting, core facility, or customer-facing work.

What This Role Is Not

Not wet lab. The work is computational.
Not sales. No quota.
Not pure research. You succeed when customers succeed.
Travel and location. Most work is virtual. Expect occasional travel to customer sites and conferences. The role is based in Bellevue, WA. No relocation assistance.