LLM Researcher and Engineer — Multimodal Large Language Models for Drug Discovery and Molecular Science
New York City, NY · Hybrid (in office Tuesday–Thursday; remote optional Monday and Friday)
The Opportunity
Large language models learn from text, but much of what matters in molecular science lives in other forms: molecular graphs, three-dimensional structures, and time-series data from simulation and experiment. Teaching a model to reason across those modalities, and training it at a scale where that reasoning becomes useful, is an open research problem. This role works on it directly, from foundational LLM research through large-scale multimodal training, with the aim of turning these models into tools that produce real scientific discoveries in drug design.
The client is a New York–based research company in computational biochemistry and drug discovery, and the team is interdisciplinary and highly collaborative. StaffRight Associates is managing this search on the client's behalf.
The Research Environment
The client combines specialized simulation hardware it designed in-house with high-performance commodity systems, machine learning, and other computational methods to model molecular behavior at the atomic level and design highly selective drug candidates. LLM work runs on the client's own large-scale computing infrastructure and feeds into both in-house programs and drug discovery efforts with outside partners.
What You'll Do
- Scale and optimize training and inference for large models on the client's specialized, large-scale computing infrastructure.
- Design pre-training data pipelines and the distributed, parallel training setups they feed.
- Develop post-training methods, including instruction tuning, reinforcement learning, and contrastive learning.
- Extend LLMs to non-text modalities such as 3D structures, molecular graphs, and time series.
- Investigate foundational questions in LLM research that bear on scientific applications.
- Partner with scientists across disciplines to apply these models to open questions in drug discovery and molecular science.
What You Bring
<>Required
- An impressive record of academic or professional achievement in machine learning.
<>Also Valued
- Deep expertise in one or more of: Large-Scale Machine Learning Systems, LLM Architecture and training, or Multimodal Learning.
- Strong Python programming skills.
- Prior exposure to Drug Discovery or Biomolecular Simulation is helpful but not expected; the client weighs curiosity, versatility, and a record of innovation in machine learning more heavily.
Paths Into This Role
Candidates whose doctoral or postdoctoral work centered on language modeling, distributed deep learning, representation learning, or geometric and graph-based machine learning will find that training applies directly here, as will researchers in Computer Science, Statistics, Physics, or Applied Mathematics who moved into large-scale model development.
Compensation & Total Rewards
- Expected annual base salary: $450,000–$1,000,000, set according to factors such as experience and education.
- Sign-on and year-end bonuses as variable pay.
- A generous benefits package with relocation and immigration support.
Joining StaffRight Associates
When you work with StaffRight Associates, you get a recruiting partner who understands the substance of your work and represents it accurately to the teams hiring for it. We focus on matching researchers with problems worth their training, and we'll give you a clear, candid picture of each opportunity before you decide to pursue it.
