X is looking to revolutionize the industrial world by making material transformation intelligent, aiming to reduce massive waste in material harvesting and processing, which is currently faced with challenges like resource exhaustion, rising energy costs, and a sizable carbon footprint. The company is building a system that combines sensing, multimodal AI, agentic digital twins, and advanced physics-based simulation to automate the continuous optimization of complex industrial processes.
Requirements
- 3+ years of experience in software engineering and applied machine learning (Python, PyTorch, or JAX).
- Experience working with Large Language Models (LLMs) and Vision-Language Models (VLMs) in applied settings, including prompt engineering or fine-tuning or RAG.
- Strong understanding of Graph data structures, Knowledge Graphs (e.g., Neo4j, NetworkX), or Graph Neural Networks (GNNs), including the handling of unstructured and/or messy real-world data such as documents, images, videos, scanned diagrams, and sensor feeds.
- Experience implementing LLM-driven code generation pipelines, specifically utilizing function calling or tool-use patterns where agents generate and execute code (e.g., Python, SQL, or Cypher) to interact with external environments or data stores.
- Experience with Agentic workflows (e.g., LangChain, AutoGen) where models perform multi-step reasoning.
- Experience with MLOps best practices, including model deployment, monitoring, and designing pipelines that allow distinct components to interoperate seamlessly.
- Background in Computer Vision and VLMs, specifically object detection or segmentation on technical imagery or diagrams.
Responsibilities
- You will design and implement state-of-the-art systems that extract structured semantic meaning from complex real world environments.
- You will solve the critical challenge of reconciling disparate data modalities, e.g., physical asset detection, sensor data, quality reporting, leveraging these inputs to build and refine models which simulate physical systems.
- You will engineer sophisticated Agentic RAG frameworks where Large Language Models reason over graph structures to perform multi-step logical deduction.
- You will tackle the engineering complexity of creating gold standard digital models from noisy real-world data.
- You will design resilient data pipelines that handle ambiguity and disparate formats at scale, ensuring reliability across documents, images, and telemetry.
- You will bridge the gap between experimental ML research, partner-oriented sprints to demonstrate value, and scalable production systems, driving the technical roadmap from initial prototype to deployed pilot.
- You will architect Agentic RAG workflows where VLMs (Vision-Language Models) and LLMs reason together to generate digital twins.
Other
- Bachelor's degree in Computer Science, AI, Engineering, or equivalent practical experience.
- A "0 to 1" mindset with the ability to thrive in ambiguity and define technical roadmaps.
- Interest in industrial automation, physics-based simulation, or AI for Science applications.
- Demonstrated ability to build self-correcting agentic loops where models iteratively write, execute, and debug code (e.g., "Code Interpreter" patterns), particularly for generating simulation logic or automating data analysis in scientific/engineering contexts.
- Familiarity with Reinforcement Learning (RL) concepts, particularly as applied to LLM post-training or optimization problems.
