AI Research Scientist
New York
Phare Health
We build AI tools to make claims more accurate and transparent, so that you can focus on patients not price tags.About Us
Our mission is to make healthcare reimbursement transparent and fair (/phare), so providers can spend more time caring for patients and less time haggling over costs. We specifically focus on the most complex AI challenges that require novel R&D, with a team that blends AI researchers and engineers with clinicians, and payment experts. Backed by General Catalyst, we’re scaling quickly - join us!
The Role
You’ll own the architecture, training, and optimization of large-scale transformer-based pipelines, wrangling PyTorch, GPUs, and distributed infrastructure to push forward SOTA. Think “research lab rigor” fused with “startup shipping speed.” Expect to:
Design and iterate on new transformer and hybrid text architectures; scale promising ideas across multi-GPU / multi-node clusters with PyTorch.
Drive the research roadmap: propose experiments, benchmark against state of the art, and publish or open-source meaningful advances.
Build retrieval-augmented generation (RAG) pipelines and lightweight agent workflows, balancing accuracy, latency, and cost.
Convert research prototypes into reliable services with CI/CD, monitoring, and rollback.
Partner with product and design to translate model capabilities into intuitive user experiences.
Who We’re Looking For
3+ years training large transformer models in Python/PyTorch at scale.
Peer-reviewed publications or significant open-source work in text modelling.
Proven end-to-end ownership: architecture → distributed training → deployment.Fluency with Lightning/FSDP, Pytorch, Hugging Face, WandB, Ray/Kubeflow, Docker
Deep expertise in text modelling; clinical-text knowledge not required.
Bonus points
RLHF or policy-optimisation methods (PPO, TRPO, DPO).
Familiarity with healthcare ontologies or claims data.
Benefits
Top-of-market compensation (salary + equity)
Flexible PTO & hybrid culture (SoHo HQ 3 days/wk; exceptional remote considered)
Mission-driven, collaborative team
Twice-yearly offsites to align, build, and celebrate.
Hiring Process
Initial application.
Intro call: Discuss your background, career goals, and our mission.
2 x Technical interviews: A programming or system design exercise focused on real-world data challenges.
Referees: We ask for 2 referees who can speak to your professional/technical work
Culture interview: Ways of working, and a chance to ask questions
Offer
* Salary range is an estimate based on our AI, ML, Data Science Salary Index 💰
Tags: Architecture CI/CD Docker FSDP GPU Kubeflow Open Source Pipelines Python PyTorch R RAG R&D Research RLHF Weights & Biases
Perks/benefits: Equity / stock options Flex hours Flex vacation Salary bonus Startup environment
More jobs like this
Explore more career opportunities
Find even more open roles below ordered by popularity of job title or skills/products/technologies used.