Meta’s Reality Labs Research (RL-R) brings together a team of researchers, developers, and engineers to create the future of Mixed Reality (MR), Augmented Reality (AR), and Wearable Artificial Intelligence (AI). The Materials and Systems Innovation (MSI) group within Reality Labs Research creates and accelerates breakthrough materials and device technologies that unblock the path to low-cost, all-day wearable AR devices and advanced sensing and actuating systems for robotics. We identify key technology gaps requiring step-change innovation, build AI-driven autonomous discovery pipelines to compress development timelines, leverage external partners to accelerate research, and deliver high-quality technology solutions through cross-functional, high-performing teams.
In this role, you will pioneer the application of generative AI to design novel compounds and molecular crystals, directly accelerating the discovery of next-generation materials for AR/VR devices and advanced robotic systems. Working at the frontier of deep generative modeling, computational chemistry, and agentic AI, you will develop and deploy state-of-the-art models — including diffusion models, flow matching, and transformer-based architectures — that predict and generate stable crystal structures and molecular candidates with target properties. Your work will be tightly integrated into our AI-driven autonomous discovery platform, collaborating with computational chemists and AI agent scientists to close the loop from molecular design to experimental validation.
Together, we are going to build advanced prototypes, technologies, and toolsets that can advance how people interact with their surroundings. We invite you to join us as we work to bring these technologies from research to reality.
Responsibilities
- Develop, train, and deploy generative models (diffusion models, flow matching, variational autoencoders, transformer-based architectures) for molecular and crystal structure generation, property-conditioned design, and crystal structure prediction (CSP)
- Design and implement reinforcement learning and alignment strategies (e.g., physics-informed reward signals from machine-learned interatomic potentials) to steer generative models toward physically stable and synthesizable candidates
- Build foundational models and scalable pretraining pipelines that unify generative and predictive learning across molecules and crystalline materials, handling both discrete atom types and continuous 3D geometries
- Collaborate closely with computational chemists to integrate first-principles calculations (DFT, force fields), molecular dynamics simulations, and domain-specific constraints into generative workflows
- Partner with AI agent scientists to embed generative molecular design capabilities into LLM-based multi-agent systems, enabling closed-loop autonomous experiment planning, candidate generation, and decision making
- Curate, preprocess, and manage large-scale molecular and crystal structure datasets for model training and benchmarking
- Establish rigorous evaluation frameworks — measuring validity, novelty, uniqueness, stability, and synthesizability of generated structures — and benchmark against state-of-the-art methods
- Contribute to the architecture and roadmap of the autonomous materials-discovery platform, ensuring generative design modules interface seamlessly with robotic workcells, characterization instruments, and data infrastructure
Minimum Qualifications
- Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience
- Ph.D. degree in Machine Learning, Computational Chemistry, Materials Science, Chemical Engineering, Physics, or a closely related technical field
- 3+ years of research experience in generative modeling applied to molecular systems, crystal structures, or materials science (academic or industry)
- Familiarity with large-scale molecular and crystal databases and data processing pipelines for chemical data
- Demonstrated expertise in deep generative models — including diffusion models, flow matching / continuous normalizing flows, variational autoencoders, or autoregressive models — with applications to 3D molecular or crystal structure generation
- Programming proficiency in Python with hands-on experience in PyTorch or JAX; proficiency in building, training, and evaluating large-scale deep learning models
- Track record of first-author publications in top-tier ML or computational chemistry venues (e.g., NeurIPS, ICML, ICLR, JACS, Nature Computational Science, Digital Discovery)
- Solid understanding of crystallography fundamentals— and molecular representations (molecular graphs, SMILES, 3D conformers) Experience integrating ML models into agentic AI frameworks or LLM-based multi-agent systems for autonomous scientific discovery
- Hands-on experience with computational chemistry tools and simulation frameworks (DFT codes such as VASP/Gaussian, molecular dynamics with LAMMPS/OpenMM/ASE, force field development)
- Experience with crystal structure prediction (CSP) pipelines, including lattice energy ranking and structure relaxation using machine-learned interatomic potentials
- Demonstrated ability to collaborate across disciplines — bridging ML research with experimental chemistry, materials science, and software engineering teams
- Experience building or fine-tuning foundation models (100M+ parameters) for chemical or materials domains, including multimodal architectures that jointly handle molecular graphs, 3D coordinates, and periodic lattice structures
- Knowledge of geometric deep learning, equivariant neural networks, or graph neural networks for molecular property prediction
- Familiarity with reinforcement learning or RLHF-style alignment techniques applied to molecular or materials generation