Advisor - Applied Deep Learning Architect
Eli Lilly and Company
2 hours ago
On-site
San Diego, CA
IT
Responsibilities:
- Deep Learning Architecture Analysis & Direction: Design, implement, and evaluate generative and predictive deep learning architectures (transformers, diffusion models, flow-matching models, graph neural networks); drive architectural decisions to improve foundational models for biologics drug discovery.
- Multi-Modal Representation Learning: Develop multi-modal embeddings unifying protein sequence, structure, and molecular fingerprints; research tokenization and fusion methods to improve generation quality and property prediction.
- Cross-Modality Molecular Modeling: Research joint modeling of proteins and small molecules for ADCs, antibody–peptide conjugates, T-cell engagers, and other multi-component biotherapeutics.
- Physics-Informed Training Objectives: Partner with MD scientists to integrate physics-based priors, molecular dynamics, and energy-aware objectives; improve physical realism and developability across modalities.
- Scientific Scouting & Knowledge Transfer: Identify high-impact AI/ML directions for the foundational model platform; educate and transfer knowledge to domain experts to enable cross-functional collaboration.
Basic Qualifications:
- Ph.D. in Computer Science, Artificial Intelligence, Theoretical Computer Science, Applied Mathematics, Computational Biology, Physics, or related field.
- Strong deep learning expertise (transformers, diffusion models, flow-matching networks, variational autoencoders, graph neural networks).
- Proficiency in Python and modern AI/ML frameworks (PyTorch or TensorFlow); familiarity with Git, code review, testing, and documentation.
Preferred Qualifications:
- 1–3 years industry experience developing/deploying novel deep learning architectures.
- Familiarity with protein engineering and biomolecular ML; experience applying ML to antibody, nanobody, or peptide design (strongly preferred).
- Experience with multi-modal architectures fusing sequence/structure/functional annotations across modalities (e.g., protein–peptide, protein–ligand, protein–small molecule).
- Experience integrating molecular dynamics, force-field representations, or physics-based priors into ML for molecular design/optimization.
- Experience with distributed training, GPU-accelerated workflows, and large-scale performant model training/inference.
- Exposure to experimental biologics workflows (phage display, yeast display, directed evolution) is a plus.
- High-impact publication record and strong oral/written communication across disciplines.
Benefits (as stated):
- Eligibility for company bonus; comprehensive benefit program including 401(k), pension, vacation, medical/dental/vision/prescription drug benefits, flexible benefits (healthcare/dependent day care FSAs), life insurance/death benefits, time off/leave of absence, and well-being benefits.
- Deep Learning Architecture Analysis & Direction: Design, implement, and evaluate generative and predictive deep learning architectures (transformers, diffusion models, flow-matching models, graph neural networks); drive architectural decisions to improve foundational models for biologics drug discovery.
- Multi-Modal Representation Learning: Develop multi-modal embeddings unifying protein sequence, structure, and molecular fingerprints; research tokenization and fusion methods to improve generation quality and property prediction.
- Cross-Modality Molecular Modeling: Research joint modeling of proteins and small molecules for ADCs, antibody–peptide conjugates, T-cell engagers, and other multi-component biotherapeutics.
- Physics-Informed Training Objectives: Partner with MD scientists to integrate physics-based priors, molecular dynamics, and energy-aware objectives; improve physical realism and developability across modalities.
- Scientific Scouting & Knowledge Transfer: Identify high-impact AI/ML directions for the foundational model platform; educate and transfer knowledge to domain experts to enable cross-functional collaboration.
Basic Qualifications:
- Ph.D. in Computer Science, Artificial Intelligence, Theoretical Computer Science, Applied Mathematics, Computational Biology, Physics, or related field.
- Strong deep learning expertise (transformers, diffusion models, flow-matching networks, variational autoencoders, graph neural networks).
- Proficiency in Python and modern AI/ML frameworks (PyTorch or TensorFlow); familiarity with Git, code review, testing, and documentation.
Preferred Qualifications:
- 1–3 years industry experience developing/deploying novel deep learning architectures.
- Familiarity with protein engineering and biomolecular ML; experience applying ML to antibody, nanobody, or peptide design (strongly preferred).
- Experience with multi-modal architectures fusing sequence/structure/functional annotations across modalities (e.g., protein–peptide, protein–ligand, protein–small molecule).
- Experience integrating molecular dynamics, force-field representations, or physics-based priors into ML for molecular design/optimization.
- Experience with distributed training, GPU-accelerated workflows, and large-scale performant model training/inference.
- Exposure to experimental biologics workflows (phage display, yeast display, directed evolution) is a plus.
- High-impact publication record and strong oral/written communication across disciplines.
Benefits (as stated):
- Eligibility for company bonus; comprehensive benefit program including 401(k), pension, vacation, medical/dental/vision/prescription drug benefits, flexible benefits (healthcare/dependent day care FSAs), life insurance/death benefits, time off/leave of absence, and well-being benefits.