E
Physical AI Engineer
Eli Lilly
3 hours ago
Full-time
On-site
Indianapolis, Indiana, United States
As a Physical AI Engineer at Eli Lilly in Indianapolis, IN, you will deploy advanced AI models into robotic, computer vision, and sensor-based systems supporting the Lilly Medicine Foundry. The focus is on improving safety, speed, and resilience through edge-to-cloud architectures, digital twin validation, and end-to-end integration into manufacturing workflows.
Role Focus
You will lead the deployment and integration of AI capabilities across robotic and sensing platforms used in real-world manufacturing environments, including cameras, AGVs, co-bots, drones, and quadruped robots. Work includes designing sensor-to-decision reference architectures, validating models through simulation and digital twins prior to GxP areas, and ensuring systems are safe, secure, compliant, and scalable. Key Responsibilities
Deploy and integrate advanced AI models, including machine learning, computer vision, generative AI and agents, and reinforcement learning, into robotic and sensing systems to improve decision-making and adaptability in manufacturing. Design, build, and operate an edge-to-cloud reference architecture: GPU edge inference on site, cloud model training on AWS, OT/IT data ingestion integrations (e.g., Ignition, MQTT, AWS IoT Greengrass, Aveva PI, Azure IoT Hub), data landing in the Foundry data platform, and orchestration supported by an ontology across layers. Assess build, buy, and hybrid options for edge hardware and edge-to-cloud software to support production deployments. Partner with engineers and data scientists to optimize AI algorithms for real-time performance on edge hardware, emphasizing minimal latency and robust behavior under changing conditions. Manage model lifecycle activities including training, versioning, performance monitoring, drift detection, and retraining. Create simulation-ready asset libraries from CAD data and the equipment registry, and use simulation platforms (e.g., NVIDIA Omniverse, Isaac Sim) to evolve from static models to dynamic digital twins, generate synthetic data for training, and test robot, co-bot, and AGV behavior and safety rules prior to deployment. Enable twin-based use cases such as lockout/tagout and SOP guidance, wayfinding to equipment, virtual tours and XR training, and smart P&ID and maintenance. Integrate vision and sensor outputs with manufacturing execution and quality systems so detections flow into an exception queue with an audit trail of Quality judgment on each AI flag, including cleaning verification and changeover checks. Expose equipment and twin data to AI agents (e.g., via MCP servers on OT data platforms) and generate incident summaries for HSE and Quality review. Develop verification methods to test and validate AI-driven robotics and vision systems in complex manufacturing environments. Design testing protocols and performance benchmarks aligned with GxP and Computer System Validation requirements in partnership with Quality. Work with safety experts to develop and implement safety mitigations and to ensure compliance with industry standards and regulations for AI-driven systems in unstructured or dynamic environments. Apply and verify safety rules for AGVs and co-bots, integrate with safety controllers, and support HSE use cases such as lockout/tagout, near-miss analysis, and early incident detection, including governance for vision-based checks. Ensure cybersecurity architecture review readiness, apply enterprise data governance and classification for image, video, and sensor data, and complete AI intake and risk assessment before deployment. Oversee deployment into production operations, moving pilots from existing spaces and labs to the Foundry site as dependencies are met (including site readiness, instrument and camera connectivity, and MES and ERP integration for AGVs), with troubleshooting and optimization during integration. Collaborate with technology and integration partners (e.g., NVIDIA, AWS, robotics, camera and drone vendors) to deliver partner-backed pilots and evaluate which partner services are production-ready for edge vision near GMP areas. Bridge Digital, Robotics, Facilities, and Technology at Lilly, and contribute to pilot selection and sequencing based on value and dependencies. Maintain documentation for system deployment processes, safety protocols, validation and testing results, and AI integration methods. Provide regular stakeholder reporting on system performance and improvements. Required Qualifications
Bachelor’s degree in Robotics, Computer Science, Electrical Engineering, Mechatronics, or a related technical field. Minimum 5 years of experience in robotics, automation, or AI-driven systems deployment in a manufacturing or industrial setting. Experience deploying computer vision or other AI models at the edge (accelerated inference, video analytics pipelines) with models trained in the cloud, ideally in a regulated or GMP manufacturing environment. Strong programming skills in Python, C++, or ROS (Robot Operating System). Knowledge of AI frameworks (TensorFlow, PyTorch, OpenCV, etc.) and model optimization and serving tools (e.g., ONNX, TensorRT, Triton). Knowledge of industrial communication protocols (e.g., OPC-UA, MQTT, Modbus, Ethernet/IP). Experience with industrial controls systems and Safety Controllers. Experience with edge AI and GPU platforms (e.g., NVIDIA Jetson, DeepStream) and edge-to-cloud architectures for vision and sensor data. Experience with cloud services for IoT, data, and machine learning (e.g., AWS IoT Greengrass, SageMaker, S3, GPU compute); familiarity with enterprise data platforms (e.g., Databricks, Microsoft Fabric). Familiarity with OT/IT data integration and historian platforms (e.g., Ignition, Aveva PI, HighByte) and manufacturing execution systems. Familiarity with simulation and digital twin tools (e.g., NVIDIA Omniverse, Isaac Sim, OpenUSD, CAD) and synthetic data generation. Familiarity with generative AI, LLMs, and agent frameworks (including MCP) applied to operations, quality, or safety use cases. Strong analytical and problem-solving skills, including troubleshooting and optimizing both hardware and software components of robotic systems. Familiarity with industry safety standards and risk mitigation strategies for robotics and automation (e.g., ISO 10218, ISO 13482, ISO 3691-4 for driverless industrial trucks and AGVs, ISO 13849, EN 61508), and OT cybersecurity practices (e.g., IEC 62443). Excellent communication and teamwork abilities, including the ability to collaborate across IT, OT, Quality, HSE, and facilities teams and work with external partners. Solid knowledge of Computer System Validation process and GxP requirements (e.g., GAMP 5, 21 CFR Part 11, EU Annex 11), including validation of AI-enabled systems. Experience in pharmaceutical or biologics manufacturing operations (e.g., changeover, cleaning verification, API processes) and integrating to quality and validation systems. Demonstrated ability to analyze, anticipate, and resolve complex issues through sound problem-solving; ability to effectively prioritize and elevate issues. Familiarity with data governance and AI risk-management practices for image, video, and sensor data. Demonstrated learning agility and curiosity. Initial Pilots and Foundry Scope
Initial pilots will run in existing space and labs before moving to the Foundry site. The new Lilly Medicine Foundry (LMF) will use the latest technology to augment the current clinical supply chain for small molecules (SM), oligonucleotides, peptides, and Antibody Drug Conjugates (ADCs), monoclonal antibodies and bioconjugates, and add new capabilities including mRNA. The facility will be built with high-tech equipment and highly integrated, automated manufacturing systems, with a focus on minimizing environmental impact. Technologies You May Work With
machine learning, computer vision, generative AI, agents, reinforcement learning, edge inference, AWS, Ignition, MQTT, AWS IoT Greengrass, Aveva PI, Azure IoT Hub, GPU edge inference on site, ontology, Python, C++, ROS (Robot Operating System), TensorFlow, PyTorch, OpenCV, ONNX, TensorRT, Triton, OPC-UA, Modbus, Ethernet/IP, NVIDIA Jetson, DeepStream, SageMaker, S3, Databricks, Microsoft Fabric, HighByte, NVIDIA Omniverse, Isaac Sim, OpenUSD, CAD, MCP, video analytics pipelines, LLMs, IEC 62443, GxP, Computer System Validation, GAMP 5, 21 CFR Part 11, EU Annex 11 Compensation and Location
Location:
Indianapolis, IN (onsite) Estimated annual compensation:
USD 118,500 - 211,200 Benefits
Company bonus (depending, in part, on company and individual performance) Eligibility to participate in a company-sponsored 401(k) Pension Vacation benefits Eligibility for medical, dental, vision, and prescription drug benefits Flexible benefits (e.g., healthcare and/or dependent day care flexible spending accounts) Life insurance and death benefits Certain time off and leave of absence benefits Well-being benefits (e.g., employee assistance program, fitness benefits, and employee clubs and activities)
#J-18808-Ljbffr
You will lead the deployment and integration of AI capabilities across robotic and sensing platforms used in real-world manufacturing environments, including cameras, AGVs, co-bots, drones, and quadruped robots. Work includes designing sensor-to-decision reference architectures, validating models through simulation and digital twins prior to GxP areas, and ensuring systems are safe, secure, compliant, and scalable. Key Responsibilities
Deploy and integrate advanced AI models, including machine learning, computer vision, generative AI and agents, and reinforcement learning, into robotic and sensing systems to improve decision-making and adaptability in manufacturing. Design, build, and operate an edge-to-cloud reference architecture: GPU edge inference on site, cloud model training on AWS, OT/IT data ingestion integrations (e.g., Ignition, MQTT, AWS IoT Greengrass, Aveva PI, Azure IoT Hub), data landing in the Foundry data platform, and orchestration supported by an ontology across layers. Assess build, buy, and hybrid options for edge hardware and edge-to-cloud software to support production deployments. Partner with engineers and data scientists to optimize AI algorithms for real-time performance on edge hardware, emphasizing minimal latency and robust behavior under changing conditions. Manage model lifecycle activities including training, versioning, performance monitoring, drift detection, and retraining. Create simulation-ready asset libraries from CAD data and the equipment registry, and use simulation platforms (e.g., NVIDIA Omniverse, Isaac Sim) to evolve from static models to dynamic digital twins, generate synthetic data for training, and test robot, co-bot, and AGV behavior and safety rules prior to deployment. Enable twin-based use cases such as lockout/tagout and SOP guidance, wayfinding to equipment, virtual tours and XR training, and smart P&ID and maintenance. Integrate vision and sensor outputs with manufacturing execution and quality systems so detections flow into an exception queue with an audit trail of Quality judgment on each AI flag, including cleaning verification and changeover checks. Expose equipment and twin data to AI agents (e.g., via MCP servers on OT data platforms) and generate incident summaries for HSE and Quality review. Develop verification methods to test and validate AI-driven robotics and vision systems in complex manufacturing environments. Design testing protocols and performance benchmarks aligned with GxP and Computer System Validation requirements in partnership with Quality. Work with safety experts to develop and implement safety mitigations and to ensure compliance with industry standards and regulations for AI-driven systems in unstructured or dynamic environments. Apply and verify safety rules for AGVs and co-bots, integrate with safety controllers, and support HSE use cases such as lockout/tagout, near-miss analysis, and early incident detection, including governance for vision-based checks. Ensure cybersecurity architecture review readiness, apply enterprise data governance and classification for image, video, and sensor data, and complete AI intake and risk assessment before deployment. Oversee deployment into production operations, moving pilots from existing spaces and labs to the Foundry site as dependencies are met (including site readiness, instrument and camera connectivity, and MES and ERP integration for AGVs), with troubleshooting and optimization during integration. Collaborate with technology and integration partners (e.g., NVIDIA, AWS, robotics, camera and drone vendors) to deliver partner-backed pilots and evaluate which partner services are production-ready for edge vision near GMP areas. Bridge Digital, Robotics, Facilities, and Technology at Lilly, and contribute to pilot selection and sequencing based on value and dependencies. Maintain documentation for system deployment processes, safety protocols, validation and testing results, and AI integration methods. Provide regular stakeholder reporting on system performance and improvements. Required Qualifications
Bachelor’s degree in Robotics, Computer Science, Electrical Engineering, Mechatronics, or a related technical field. Minimum 5 years of experience in robotics, automation, or AI-driven systems deployment in a manufacturing or industrial setting. Experience deploying computer vision or other AI models at the edge (accelerated inference, video analytics pipelines) with models trained in the cloud, ideally in a regulated or GMP manufacturing environment. Strong programming skills in Python, C++, or ROS (Robot Operating System). Knowledge of AI frameworks (TensorFlow, PyTorch, OpenCV, etc.) and model optimization and serving tools (e.g., ONNX, TensorRT, Triton). Knowledge of industrial communication protocols (e.g., OPC-UA, MQTT, Modbus, Ethernet/IP). Experience with industrial controls systems and Safety Controllers. Experience with edge AI and GPU platforms (e.g., NVIDIA Jetson, DeepStream) and edge-to-cloud architectures for vision and sensor data. Experience with cloud services for IoT, data, and machine learning (e.g., AWS IoT Greengrass, SageMaker, S3, GPU compute); familiarity with enterprise data platforms (e.g., Databricks, Microsoft Fabric). Familiarity with OT/IT data integration and historian platforms (e.g., Ignition, Aveva PI, HighByte) and manufacturing execution systems. Familiarity with simulation and digital twin tools (e.g., NVIDIA Omniverse, Isaac Sim, OpenUSD, CAD) and synthetic data generation. Familiarity with generative AI, LLMs, and agent frameworks (including MCP) applied to operations, quality, or safety use cases. Strong analytical and problem-solving skills, including troubleshooting and optimizing both hardware and software components of robotic systems. Familiarity with industry safety standards and risk mitigation strategies for robotics and automation (e.g., ISO 10218, ISO 13482, ISO 3691-4 for driverless industrial trucks and AGVs, ISO 13849, EN 61508), and OT cybersecurity practices (e.g., IEC 62443). Excellent communication and teamwork abilities, including the ability to collaborate across IT, OT, Quality, HSE, and facilities teams and work with external partners. Solid knowledge of Computer System Validation process and GxP requirements (e.g., GAMP 5, 21 CFR Part 11, EU Annex 11), including validation of AI-enabled systems. Experience in pharmaceutical or biologics manufacturing operations (e.g., changeover, cleaning verification, API processes) and integrating to quality and validation systems. Demonstrated ability to analyze, anticipate, and resolve complex issues through sound problem-solving; ability to effectively prioritize and elevate issues. Familiarity with data governance and AI risk-management practices for image, video, and sensor data. Demonstrated learning agility and curiosity. Initial Pilots and Foundry Scope
Initial pilots will run in existing space and labs before moving to the Foundry site. The new Lilly Medicine Foundry (LMF) will use the latest technology to augment the current clinical supply chain for small molecules (SM), oligonucleotides, peptides, and Antibody Drug Conjugates (ADCs), monoclonal antibodies and bioconjugates, and add new capabilities including mRNA. The facility will be built with high-tech equipment and highly integrated, automated manufacturing systems, with a focus on minimizing environmental impact. Technologies You May Work With
machine learning, computer vision, generative AI, agents, reinforcement learning, edge inference, AWS, Ignition, MQTT, AWS IoT Greengrass, Aveva PI, Azure IoT Hub, GPU edge inference on site, ontology, Python, C++, ROS (Robot Operating System), TensorFlow, PyTorch, OpenCV, ONNX, TensorRT, Triton, OPC-UA, Modbus, Ethernet/IP, NVIDIA Jetson, DeepStream, SageMaker, S3, Databricks, Microsoft Fabric, HighByte, NVIDIA Omniverse, Isaac Sim, OpenUSD, CAD, MCP, video analytics pipelines, LLMs, IEC 62443, GxP, Computer System Validation, GAMP 5, 21 CFR Part 11, EU Annex 11 Compensation and Location
Location:
Indianapolis, IN (onsite) Estimated annual compensation:
USD 118,500 - 211,200 Benefits
Company bonus (depending, in part, on company and individual performance) Eligibility to participate in a company-sponsored 401(k) Pension Vacation benefits Eligibility for medical, dental, vision, and prescription drug benefits Flexible benefits (e.g., healthcare and/or dependent day care flexible spending accounts) Life insurance and death benefits Certain time off and leave of absence benefits Well-being benefits (e.g., employee assistance program, fitness benefits, and employee clubs and activities)
#J-18808-Ljbffr