Senior Machine Learning Scientist (Multimodal Drug Discovery)

Coders Connect is collaborating with a cutting-edge biotech startup redefining how foundational biology meets AI. Their mission? Train multimodal foundation models from real human biological data to revolutionize drug discovery.

What You’ll Be Doing:

We’re seeking a Senior Machine Learning Scientist with a deep interest in applying state-of-the-art ML techniques—like transformers and diffusion models—to problems at the frontier of computational biology.

• Design and develop multimodal foundation models for gene regulation and drug response
  

• Integrate chemical structure, protein sequences, and single-cell transcriptomics into unified models
  

• Adopt and implement latest advances in deep learning (e.g., self-supervised learning, FSDP, dMoE)
  

• Work in cross-functional teams with biologists, engineers, and data scientists
  

• Drive hypothesis generation using ML and contribute to novel biological insights
  

Requirements

• PhD or equivalent practical experience in a technical/ML-focused field
  

• Proven experience with deep learning (transformers, GNNs, SSMs, diffusion models, etc.)
  

• Strong skills in PyTorch, JAX, or TensorFlow, and scientific libraries like NumPy, Pandas
  

• Motivation to apply ML to real-world biological or chemical datasets
  

• Bias toward rapid prototyping and practical outcomes

Bonus Points For:

• Prior work in computational biology or drug discovery
  

• Experience with contrastive/multimodal/self-supervised learning
  

• Familiarity with large-scale distributed training and GPU optimizations

The Stack:

• Python, PyTorch/JAX/TensorFlow
  

• Large-scale ML toolkits (e.g., flash attention, FSDP)
  

• Transcriptomics, protein sequence data, chemical structure modeling
  

Benefits

• Unlimited PTO
  
  
• Monthly lunch allowance + remote office setup stipend
  
  
• Premium medical, dental, and vision coverage for employees and dependents
  
  
• Hybrid flexibility from South San Francisco or Toronto