Your Impact at LILA

As a Machine Learning Scientist focused on Scientific Reasoning, you will help pioneer the next generation of AI systems capable of reasoning like a scientist. You’ll design novel frameworks that push the boundaries of LLM-based reasoning methods — while also implementing scalable frameworks that integrate with Lila’s platforms. This role bridges deep theoretical thinking with practical ML engineering, enabling breakthroughs in how scientific hypotheses are generated, tested, deployed and optimized.

What You'll Be Building

• Design and formalize frameworks for scientific reasoning with LLMs, including structured prompting, reasoning chains, and test-time compute.
• Explore and implement methods for in-context learning, self-reflection, and adaptive reasoning in scientific discovery workflows.
• Build scalable model prototypes that can be deployed to solve frontier scientific problems.
• Collaborate with scientists and engineers to encode domain knowledge into reasoning systems that integrate symbolic and statistical approaches.

What You’ll Need to Succeed

• PhD (preferred) or equivalent research/industry experience in Computer Science, Machine Learning, AI, Engineering, Materials Science or related fields.
• Strong programming skills in Python with deep expertise in LLM frameworks (PyTorch, HuggingFace Transformers, LangChain, LlamaIndex, and related toolkits).
• Expertise in LLM reasoning methods: in-context learning, test-time compute, chain-of-thought, or tool-augmented reasoning.
• Ability to balance theoretical research with practical ML engineering to deliver scalable solutions.

Bonus Points For

• Research experience in causal reasoning, symbolic AI, or probabilistic programming.
• Contributions to open-source LLM reasoning frameworks.
• Familiarity with scientific discovery pipelines in chemistry, biology, or materials science.
• Experience with multimodal reasoning (e.g., combining text, image, and experimental data).
• Publications in top ML/AI conferences (NeurIPS, ICML, ICLR, ACL).

Your Impact at LILA

As a member of our team in the Physical Sciences division, you will shape the ML- and agent-driven infrastructure that enables Lila’s scientific superintelligence to autonomously construct, execute, and interpret complex physics simulations. Your work will focus on building the core systems that allow AI agents to reason over and control computational experiments, spanning from electronic structure calculations to surrogate-driven atomistic modeling, and beyond. Your work will directly influence how autonomous computational processes will explore chemical and materials landscapes with unprecedented autonomy and robustness.

What You'll Be Building

• Architect and implement agentic frameworks that support dynamic, multi-stage simulation workflows for scientific tasks
• Develop pipelines enabling agents to autonomously plan, schedule, execute, and interpret computational tasks at scale
• Build integration layers and APIs that connect ML models, large-scale simulation engines, databases, and heterogeneous compute platforms.
• Work with AI researchers to productionize agent behaviors, including tool-use strategies, simulation-aware decision loops, and adaptive task planning.
• Improve the robustness, modularity, performance, and reproducibility of agent-driven computational workflows; build internal tooling for debugging, observability, and validation.

What You’ll Need to Succeed

• MS/PhD or equivalent experience in Computer Science, ML/AI, Scientific Computing, or a related technical field.
• Strong experience building ML-driven pipelines, workflow systems, or tool-use frameworks, ideally for complex or scientific applications.
• Proficiency in Python and ML ecosystems; experience with one or more compiled languages (e.g., C++, Rust, Julia) is beneficial.
• Familiarity with large-scale scientific or engineering software, including integrating external tools into automated computational workflows.
• Experience with distributed systems, HPC environments, cloud platforms, or accelerator-based computing.
• Deep understanding of modern ML architectures and their deployment in production systems (e.g., GNNs, transformers, diffusion models, multimodal or tool-using models).
• Strong engineering fundamentals: reproducibility, testing, modular design, CI/CD, and scalable ML operations.

Bonus Points For

• Experience developing or integrating agentic frameworks, autonomous ML pipelines, or multi-step tool-using agents, particularly for scientific applications
• Background in large-scale simulation frameworks, scientific workflow orchestration, or automated computational experiment platforms.
• Contributions to open-source ML infrastructure, workflow engines, agent frameworks, or scientific software ecosystems.
• Familiarity with data-centric engineering practices: streaming systems, provenance tracking, distributed metadata services, or large-scale data orchestration.
• Experience with workflow orchestration systems (e.g., Flyte, Prefect, Airflow) and container orchestration (Kubernetes).
• Familiarity with materials science simulation codes (VASP, LAMMPS) or workflow frameworks (atomate, aiida) is a plus but not required.
• Expertise in advanced ML techniques relevant to agentic reasoning (planning models, self-improving systems, multi-tool agents, RLHF/RLAIF workflows).

Your Impact at LILA

Your role in our Physical Sciences division will focus on designing and implementing state-of-the-art simulation approaches to model transport, kinetics, rare events, and reaction networks, and integrating them with AI-driven platforms for materials discovery. Your work will be integral to our efforts in predicting, designing, and controlling the behavior of complex materials and molecular systems, and their acceleration via agentic AI. You will partner with diverse teams at Lila, including machine learning experts working on scientific superintelligence and materials science experts performing real-world experiments.

What You'll Be Building

• Develop and extend molecular dynamics and Monte Carlo algorithms to capture rare events, non-equilibrium processes, transport phenomena, and mapping complex reaction networks.
• Build scalable simulation workflows that integrate statistical mechanics methods with machine learned interatomic potentials and agentic AI frameworks.
• Design methods for coupling dynamics simulations with experimental observables to enable closed-loop verification and discovery with automated labs.
• Collaborate with computational scientists, machine learning experts, and platform engineers to advance the fidelity and scalability of simulation-driven materials discovery.
• Establish reproducible, modular software pipelines for statistical mechanics and dynamics simulations that can be deployed on HPC and cloud-based infrastructure.

What You’ll Need to Succeed

• PhD or equivalent research/industry experience in Physics, Chemistry, Chemical Engineering, Mechanical Engineering, Applied Mathematics, or related fields.
• Strong background in statistical mechanics, free energy calculations, reaction mapping, non-equilibrium dynamics, and rare-event sampling.
• Demonstrated expertise with molecular dynamics, Monte Carlo, and/or kinetic simulation software and frameworks (LAMMPS, GROMACS, OpenMM, HOOMD, etc.).
• Solid programming skills and experience with scientific computing (Python, C/C++, MPI, CUDA, etc.).
• Experience running and automating simulations on HPC and/or cloud environments at scale.

Bonus Points For

• Strong publication record applying advanced statistical mechanics or dynamics simulations to molecular and materials systems, including but not limited to molecular/biomolecular systems and solid-state materials and interfaces.
• Prior work in coupling dynamics simulations with data-driven, AI-based, and/or agentic frameworks.
• Good familiarity with machine learning frameworks (PyTorch, JAX, TensorFlow, etc.)
• Prior experience working with machine learned interatomic potentials, including model training, fine-tuning, and data generation
• Worked closely with experimental teams to extract and corroborate experimental observables from dynamics simulations

Your Impact at LILA

Your role in our Physical Sciences division will center on developing the next generation of in silico materials discovery methods, from creating autonomous workflows and data-driven pipelines to building the interface between simulation and AI. You’ll pioneer strategies that enable agents to reason over simulation data, extract latent insights, and guide hypothesis generation and materials design. Your work will expand how we leverage simulation outputs for discovery, accelerating the integration of physics-based modeling and AI reasoning systems. You’ll collaborate with experts in areas spanning simulation, AI agents, and experimental automation to push the boundaries of digital discovery.

What You'll Be Building

• Develop methods and workflows for in silico materials discovery that connect physics-based simulations, generative models, and agentic AI systems.
• Build intelligent pipelines where AI agents can design, launch, interpret, and refine simulations autonomously.
• Design frameworks that utilize simulation data more effectively for prediction, inference, and discovery, including automatic feature extraction, model training, and data-driven exploration.
• Prototype and evaluate new paradigms for simulation-aware agents that can learn from and act on scientific simulations.
• Design data representations, metadata standards, and APIs that enable seamless flow of information between simulations, machine learning models, and experimental databases.
• Create scalable, modular workflows that bridge electronic structure, atomistic, and mesoscale simulations with AI-driven reasoning and hypothesis generation.
• Collaborate with computational scientists, machine learning experts, and platform engineers to integrate in silico discovery pipelines into Lila’s broader scientific superintelligence ecosystem.

What You’ll Need to Succeed

• PhD or equivalent experience in Computer Science, Materials Science, Chemistry, Physics, Applied Mathematics, or related disciplines.
• Strong foundation in in silico materials discovery, computational materials modeling, and/or simulation workflow design.
• Familiarity with large language models and their application in scientific domains, and
• Experience building AI-driven or agentic workflows for scientific automation and discovery.
• Solid programming skills in Python and scientific computing frameworks
• Familiarity with atomistic simulation software and libraries (e.g., VASP, LAMMPS, ASE, Pymatgen, etc.).

Bonus Points For

• Strong publication record in in silico materials discovery, simulation-AI integration, AI-driven inverse design.
• Familiarity with scientific agent architectures, large-scale reasoning systems, or multi-agent frameworks for hypothesis generation and experimental planning.
• Familiarity with ontologies, metadata standards, and data infrastructure for scientific simulations.
• Experience with automated experiment–simulation loops, integrating computational predictions with robotic or cloud-based laboratory platforms.

Your Impact at LILA

As a ML Research Scientist - Multimodal Data Extraction, you will advance Lila’s vision of scientific superintelligence by developing foundation models that autonomously read, interpret, and structure scientific knowledge across text, images, and experimental data in the physical sciences. Your research will help unify the world’s scientific information into machine-understandable form, powering reasoning, prediction, and autonomous discovery across materials science and chemistry.

What You'll Be Building

• Research and develop AI systems that extract and structure knowledge from diverse scientific sources.
• Design and fine-tune large language, multi-modal and specialized models for factual, interpretable data extraction.
• Build scalable pipelines for unstructured and heterogeneous scientific data, integrating text, tables, and visuals.
• Collaborate with domain experts to align extracted data with real-world discovery workflows.
• Publish research that advances the state of the art in multimodal understanding and AI-driven knowledge extraction.

What You’ll Need to Succeed

• PhD (or equivalent research experience) in Computer Science, Chemistry, Materials Science, or related field.
• Expertise in machine learning, NLP, and vision–language modeling using PyTorch and Hugging Face Transformers.
• Proven ability to train, fine-tune, and evaluate LLMs and multimodal models for scientific data extraction.
• Strong understanding of data structures and representations used in the physical sciences.
• Demonstrated research impact through publications, preprints, or open-source work (e.g., NeurIPS, ICLR, ICML, ACL, EMNLP, Scientific Journals).

Bonus Points For

• Experience with multimodal fusion architectures and document-level understanding.
• Knowledge of scientific document parsing (OCR, table extraction, figure-caption linking).
• Familiarity with knowledge graph construction or reasoning systems for science.
• Experience with noisy or heterogeneous real-world scientific data.
• Collaborative mindset and passion for advancing AI in the physical sciences.

Your Impact at LILA

As a member of our team in the Physical Sciences division, you will design and build intelligent agent-driven systems that enable AI-accelerated and AI-orchestrated process engineering across a broad range of industrial applications. The core mission of this role is to develop methods by which AI agents can reason about, design, simulate, optimize, and operate complex physical and chemical processes using existing or ML-driven process engineering tools. You will focus on creating agentic infrastructures that allow AI systems to plan and execute multi-step process engineering workflows, ranging from process synthesis and flowsheet generation to steady-state and dynamic simulation, control strategy design, and techno-economic evaluation. Your work will directly shape how Lila’s scientific superintelligence performs closed-loop autonomous process engineering to solve real-world problems.

What You'll Be Building

• Architect and implement agentic frameworks that support end-to-end process engineering workflows, including process setup, simulation, optimization, and analysis.
• Develop AI agents capable of autonomously planning, executing, and iterating on process engineering tasks using existing tools (e.g., steady-state and dynamic simulators, optimizers, and data systems).
• Explore agentic approaches for advanced tasks such as process intensification, control co-design, real-time optimization, and closed-loop learning from operational data.
• Improve robustness, interpretability, and reproducibility of agent-driven process engineering workflows; build internal tooling for debugging, observability, validation, and auditability.
• Work with interdisciplinary teams to apply agentic process engineering to a broad range of industrial applications

What You’ll Need to Succeed

• PhD or equivalent experience in Chemical Engineering, Industrial Engineering, Systems Engineering, or a closely related field.
• Research experience in method development for process engineering, a strong publication record in this area or established industry experience
• Hands-on experience with process simulation and optimization tools (commercial or open-source), including steady-state and dynamic modeling.
• Proficiency in Python and scientific/engineering computing ecosystems
• Experience integrating external engineering tools or simulators into automated workflows via APIs, scripting interfaces, or custom wrappers.
• Familiarity with distributed systems, HPC environments, cloud platforms, or scalable compute infrastructure.

Bonus Points For

• Experience developing or integrating agentic frameworks, autonomous planners, or multi-step tool-using AI systems for engineering or scientific domains.
• Experience building computational pipelines, automation systems, or tool-use frameworks for complex engineering or scientific workflows.
• Experience with digital twins, real-time optimization, or model-predictive control frameworks.
• Background in techno-economic analysis (TEA), life-cycle assessment (LCA), or sustainability-driven process design.
• Contributions to open-source engineering software, ML infrastructure, workflow engines, or agent frameworks.

AI Resident – 2026 Cohort

The AI Residency Program is a full-time research opportunity designed to bridge the gap between academic research and industry applications in AI for materials science. Residents will work closely with Lila scientists and engineers on high-impact, open-science projects, with the option to focus on either fundamental or applied research.

• Duration: 6–12 months (extension possible)
• Start Dates: First hires beginning January 2026, with rolling applications and additional intakes in Summer and Fall 2026
• Cohort Size: Small group of selected residents
• Mentorship: Pairing with technical mentors, feedback from cross-functional teams
• Resources: Access to proprietary datasets, high-performance compute, and Lila’s research infrastructure

Research areas include ML-accelerated simulations, Bayesian methods, representation learning, generative models, agentic science, and ML-driven automation.

Your Impact at Lila

The Lila Sciences AI Residency is a full-time research program at the intersection of artificial intelligence and materials science. As a resident, you'll join a cohort of researchers tackling open-ended scientific challenges alongside Lila’s world-class team of scientists and engineers. With access to proprietary datasets, high-performance compute infrastructure, and experienced mentors, you'll pursue ambitious research projects with both academic and real-world impact. Publishing is encouraged but not required — what matters most is pushing the frontier of scientific discovery.

What You'll Be Building

• Design and execute independent research projects in AI for materials science
• Collaborate with Lila scientists and engineers on cutting-edge, open-science initiatives
• Explore domains such as ML-accelerated simulations, Bayesian methods, representation learning, generative AI, agentic science, and ML-driven automation
• Contribute to collaborative team research and co-develop novel approaches to scientific discovery
• Share findings internally and externally; publications are welcome but not mandatory

What You’ll Need to Succeed

• Degree in Materials Science, Chemistry, Computer Science, AI/ML, Physics, Mathematics, or related field (Bachelor’s, Master’s, or PhD)
• Proficiency in Python and deep learning frameworks (e.g., PyTorch)
• Experience working with large-scale datasets or simulations
• Familiarity with modern AI/ML architectures and training techniques
• Strong research background, demonstrated through publications, thesis work, or open-source projects

Bonus Points For

• Prior work on ML applications in scientific domains (e.g., materials discovery, chemistry, simulations)
• Familiarity with Bayesian optimization, active learning, or generative models
• Experience in reinforcement learning or agent-based approaches to scientific reasoning
• Open-source contributions or collaborative research experience
• Strong communication and writing skills, especially for conveying complex scientific ideas