Job Details

Primary Work Address: 19700 Helix Drive, Ashburn, VA, 20147

TLDR: Build AI methods for 3D particle detection and structural analysis in cryo-electron tomography data, applied to chromatin organization and synaptic molecular targets.

Please include a cover letter with your application. Describe a deep learning project you have executed, ideally involving 3D image analysis, inverse problems, or physics-informed modeling. Cryo-EM/ET and computational structural biology projects are especially relevant. Discuss results, limitations, and challenges encountered. If the project was collaborative, describe your specific contributions. Include links to relevant code repositories and your GitHub/Gitlab profile, personal website, or similar evidence.

About the role:

AI@HHMI: HHMI is investing $500 million over the next 10 years to support AI-driven projects and to embed AI systems throughout every stage of the scientific process in labs across HHMI. This role is part of the AI+CryoET project within AI@HHMI, a multi-institutional project at the intersection of cryo-electron tomography (cryoET), molecular dynamics simulation, and machine learning. The project aims to develop AI methods for mesoscale structural biology, understanding how cellular macromolecules organize into higher-order structures. You will work in a team at Janelia, with experimental and computational collaborators across the Rosen lab (UT Southwestern Medical Center/HHMI), Gouaux lab (Oregon Health and Science University/HHMI), Collepardo-Guevara lab (University of Cambridge), and Villa lab (UC San Diego/HHMI).

You will develop machine learning methods for particle detection, localization, and structural analysis in cryoET data, with two interconnected aims: (1) detecting gold nanoparticle (AuNP) probes to improve reconstruction quality and identify molecular targets; (2) identifying the arrangement and connectivity of nucleosomes in chromatin that give rise to chromosome structure in cell nuclei and biochemical reconstitutions. This involves developing supervised and self-supervised AI models based on simulated as well as annotated experimental cryoET data, informed by molecular dynamics simulations of relevant biological structures.

What we provide:

• A competitive compensation package with comprehensive health and welfare benefits.
• A supportive team environment that promotes collaboration and knowledge sharing.
• Access to world-class computational infrastructure, GPU-based computing environments, and unique high-quality cryoET datasets.
• The opportunity to work directly with leading structural biologists, cryoET experimentalists, and molecular dynamics experts on a highly interdisciplinary project.
• The opportunity to engage with world-class researchers, software engineers, and AI/ML experts, contribute to impactful science, and be part of a dynamic community committed to advancing humanity's understanding of fundamental scientific questions.
• Amenities that enhance work-life balance, such as on-site childcare, free gyms, available on-campus housing, social and dining spaces, and convenient shuttle bus service to Janelia from the Washington, D.C. metro area.
• Opportunity to partner with frontier AI labs on scientific applications of AI.

What you'll do:

• Develop and evaluate deep learning models for detecting and localizing gold nanoparticles and macromolecular particles (e.g., nucleosomes, synaptic receptors) in cryoET data, and for identification of nucleosome arrangement and connectivity in chromatin.
• Develop methods to leverage gold nanoparticle detections to improve tomogram reconstruction, addressing challenges in tilt-series alignment, deformations, and low signal-to-noise conditions.
• Design and execute rigorous AI model training and evaluation pipelines, including proper handling of missing wedge artifacts, CTF effects, and sim-to-real transfer from MD-derived synthetic training data.
• Identify where additional human annotation and proofreading will be most helpful and design and guide annotation efforts.
• Contribute to scientific publications, present findings at conferences, and maintain a well-documented codebase enabling seamless reproduction and extension of results.
• Collaborate with interdisciplinary teams across multiple institutions.

What you bring:

• Master's or PhD in Computer Science, Applied Mathematics, Physics, Computational Chemistry, or a related field, or equivalent combination of education and experience.
• 3+ years training and evaluating deep learning models, particularly on 3D or volumetric image data. Experience with detection, segmentation, or inverse problems in imaging is strongly preferred.
• Strong Python skills, and proficiency in PyTorch and/or JAX. Ability to reason about neural network behavior from first principles: how architectural choices, regularization, and training procedures affect model behavior.
• Rigorous experimental design: model comparisons, ablation studies, reproducibility.
• Commitment to open science.
• Experience with scalable GPU-based computing environments on Linux HPC clusters and high-throughput processing for large-scale data.
• Excellent communication skills and keen interest in working in a truly interdisciplinary environment.

Ways to stand out:

• Experience with cryo-EM/ET data processing, tomographic reconstruction, or related inverse problems in imaging.
• Familiarity with molecular dynamics simulations (e.g., OpenMM, LAMMPS) and/or synthetic data generation for training ML models.
• Experience with differentiable rendering, neural radiance fields, or analysis-by-synthesis approaches for 3D reconstruction.
• Knowledge of cryoET software tools (IMOD, Warp, RELION, AreTomo etc.) or microscopy data formats (MRC, Zarr).
• Experience with template matching, sub-tomogram averaging, or particle picking in cryo-EM/ET contexts.