Founding ML Research Scientist

CT19, incorporated; New York (NY); On-site; USD 180,000 - 260,000; Full time

Job summary

CT19, incorporated is seeking a Founding Research Scientist to lead modelling across live pharma and biotech projects. Turn scientific questions into model tasks, build training/evaluation plans, and deliver actionable findings for research teams.

This hands-on role offers substantial ownership, working closely with founders to decide which biological problems to tackle, data to generate, and how results become reusable capabilities.

Qualifications

• Direct experience training or substantially improving large models in biology or related domains.
• Strong judgement on representations, architectures, training objectives and evaluation.
• Hands-on experience with biological datasets and issues like noise and batch effects.

Responsibilities

• Lead modelling projects end to end: define questions, select data, train, evaluate and present results.
• Train and adapt models using diverse biological data including single-cell and imaging data.
• Develop models predicting cellular states and responses, including perturbations.
• Design rigorous evaluations with held-out experiments and independent cohorts.
• Collaborate with scientists to identify valuable data to acquire or generate next.
• Translate research into reusable model capabilities with engineering teams for scale.
• Explain results and modelling decisions clearly to ML and biology teams.

Skills

Large-model training

Biology data handling

Python

PyTorch or JAX

Multi-GPU experiments

Model evaluation

Perturbation models

Cellular imaging

Cross-species translation

PhD/postdoc experience

Education

PhD or equivalent research track record

Tools

PyTorch

JAX

Job description

CT19 are partnering with an early-stage start up who are hiring a Founding Research Scientist to lead modelling across live pharma and biotech projects. You will turn scientific questions into model tasks, build the training and evaluation approach, and deliver findings that research teams can use.

This is a hands-on role with substantial ownership. You will work closely with the founders and help decide which biological problems to tackle, what data to generate, and how research results become reusable capabilities.

What you’ll do:

• Lead modelling projects end to end, from defining the scientific question and selecting datasets through training, evaluation and presentation of results.
• Train and adapt models using single-cell, perturbation, spatial, multi-omic, imaging and clinical assay data.
• Develop models that predict cellular states and responses, including responses to genetic or chemical perturbations and how results transfer across biological settings.
• Design rigorous evaluations using held-out experiments, unseen cell types or tissues, independent cohorts and, where possible, experimental validation.
• Work with scientific collaborators to identify the most valuable data to acquire or generate next.
• Translate successful research into reusable model capabilities, working alongside engineering teams to make them reliable at scale.
• Explain results and modelling decisions clearly to both ML specialists and biology teams.

What we’re looking for:

• Direct experience training or substantially improving large models, whether in biology, vision, language, multimodal learning, speech or another domain.
• Strong judgement on representations, model architecture, training objectives, fine-tuning and evaluation.
• Hands-on experience with biological datasets and an understanding of issues such as noise, batch effects, confounding and experimental controls.
• Fluency in Python and PyTorch or JAX, with experience running experiments on multi-GPU infrastructure.
• The ability to assess whether a prediction is biologically plausible and useful to a research team.
• Experience taking an ambiguous research question through to a result that others can act on.
• Experience with perturbation models, cellular imaging, cross-species translation, pharma collaborations or scientific LLM post-training would be particularly relevant.
• A PhD or equivalent research track record is welcome, though demonstrated research and modelling ability matter most.