Advisor - Agent Research

Scorpion Therapeutics; San Diego (CA); On-site; USD 152,000 - 222,000; Full time

Benefits offered by this job

401(k) and pension

Medical/dental/vision

Paid time off

Well-being benefits

Life insurance

Job summary

Scorpion Therapeutics seeks a PhD-level ML scientist to design the learning layer of a scientific agent platform, translating wet-lab and computational endpoints into trainable signals. You will enable agents to improve from experimental feedback and collaborate with scientists on molecule discovery tasks.

You will design RL environments, engineer reward functions, and post-train domain models on chemistry and biology tasks, integrating policies with RDKit and ELN/LIMS APIs, and building

Qualifications

• PhD or MS/BS with wet-lab collaboration experience.
• 1–2 years applying AI/ML in scientific disciplines.
• Hands-on experience training/post-training AI models.

Responsibilities

• Design the learning layer of a scientific agent platform for improved experiments.
• Partner with scientists to build autonomous agents for molecule discovery tasks.
• Design and build RL environments with proper state/action semantics.
• Curate reward functions from noisy scientific signals.
• Post-train domain models on chemistry and biology tasks.
• Integrate learned policies with RDKit, ML tools, and LIMS APIs to execute DMTA tasks.
• Build evaluation infra: task suites, scoring, and experiment tracking (MLflow).
• Represent Frontier AI in the research community through publications and talks.

Skills

Python

PyTorch

RL (PPO, DPO)

Molecular representation learning

Cloud & Kubernetes

Education

PhD in ML / Bioinformatics / Cheminformatics

MS + 3 yrs / BS + 5 yrs equivalent

Tools

RDKit

ELN/LIMS APIs

MLflow

HuggingFace

Job description

Position Summary

• Design the learning layer of a scientific agent platform to enable agents to improve from experimental feedback; translate wet-lab and computational endpoints into trainable signals.

Responsibilities

• Partner with scientists to build autonomous agents for molecule discovery tasks.
• Design and build reinforcement learning (RL) environments with appropriate state/action/termination semantics.
• Curate and engineer reward functions from noisy scientific signals.
• Post-train domain models (SFT, DPO/GRPO/PPO, reward modeling, distillation) on chemistry and biology tasks.
• Integrate learned policies with domain tools (RDKit, molecular graph ML, ELN/LIMS APIs, instrument drivers) to execute DMTA tasks.
• Build evaluation infrastructure: task suites, scoring harnesses, regression and experiment tracking (e.g., MLflow).
• Represent Frontier AI in the AI research community (publish, talks, reviews) and scout trends.
• Evaluate external vendors, open-source projects, and academic collaborations.

Basic Qualifications

• PhD (or MS + 3 yrs / BS + 5 yrs equivalent) in Machine Learning, Bioinformatics, Cheminformatics, Computer Science, or related field with demonstrated wet-lab collaboration/hands‑on experience.
• ~1–2 years applying AI/ML in scientific disciplines (industry postdoc counts).
• Hands‑on experience training/post‑training AI models.

Additional Preferences

• Python; deep learning frameworks (PyTorch/TensorFlow/JAX/HuggingFace).
• RL/post‑training methods (PPO, GRPO, DPO, reward modeling, RLHF/RLAIF) and libraries (TRL/verl or equivalent).
• Molecular representation learning, generative chemistry, or protein/nucleic acid models.
• Agentic AI systems (OpenAI/Anthropic Agent SDK, Langchain, Smol agents).
• Cloud end‑to‑end systems; AWS/Azure; Nextflow/Argo on Kubernetes.
• Publications in relevant ML/NLP venues; mentoring experience.

Application/Benefits

• None explicitly included beyond compensation/benefits: $151,500–$222,200 + potential company bonus; comprehensive benefits including 401(k), pension, vacation, medical/dental/vision/prescription, flexible benefits, life insurance, time off/leave, and well‑being benefits.