Offline Analytics & On-Premises ML Application Lead

Job Title: Offline Analytics & On-Premises ML Application Lead Reports To: Head of Data Analytics EV Department: Data Analytics & Connected Vehicle Intelligence Role Overview As the Offline Analytics & On-Premises ML Application Lead, you will bridge the gap between heavy industrial data engineering and cutting-edge machine learning application development. While real-time telemetry flows to the cloud, the massive, historical "cold" datasets—including high-frequency raw CAN logs, battery lifecycle testing data, and firmware diagnostic dumps—require secure, high-compute, localized management. You will own the architecture of our on-premises data infrastructure, optimize it for training complex AI/ML models (such as deep battery degradation physics and predictive maintenance systems), and build internal applications that allow R&D engineers and data scientists to build, test, and deploy AI models securely. Key Responsibilities On-Premises Data Infrastructure: Design, scale, and manage our on-premises data clusters (e.g., localized Hadoop, Spark, or high-performance NVMe time-series storage) optimized to store and process terabytes of historical EV telemetry. Offline AI/ML Application Development: Lead the development of internal software tools and web applications (using Python, Node.js, or Go) that expose machine learning models to business units, enabling automated diagnostic reporting, fleet health simulations, and warranty forecasting. Heavy Compute ML Pipelines: Architect and maintain local GPU-accelerated computing infrastructure (e.g., NVIDIA DGX systems or custom bare-metal clusters) for offline deep learning, battery cell chemistry modeling, and training resource-intensive neural networks. Data Ingestion & Synchronization: Establish efficient, secure data pipelines to ingest massive offline datasets—such as end-of-line manufacturing tests, laboratory battery cycling data, and physical workshop diagnostic downloads—and selectively sync key parameters with cloud environments. MLOps & Model Lifecycle: Implement on-premises MLOps frameworks (e.g., local MLflow, Kubeflow, or Dockerized registries) to manage version control, training datasets, and artifact storage for offline-trained models. Data Governance & Security: Enforce strict data governance, localization protocols, and access controls for sensitive intellectual property, including proprietary battery chemistry logs and vehicle performance benchmarks. Required Qualifications & Skills Experience: 6+ years of experience in data engineering, machine learning engineering, or software development, with at least 2+ years leading technical teams or complex architecture projects. Infrastructure Expertise: Deep experience building and managing on-premises distributed storage and compute frameworks (e.g., Apache Spark, HDFS, Ceph, MinIO) and containerized workflows using Docker and Kubernetes . AI/ML Application Tooling: Proven track record of building production-grade application backends (Python, Go, or Java) and integrating them with ML frameworks ( TensorFlow, PyTorch, Scikit-Learn ). Hardware & Compute Fluency: Strong understanding of bare-metal Linux administration, GPU optimization (CUDA layout), and network-attached storage (NAS/SAN) configurations for high-throughput data processing. EV & Signal Processing Knowledge: Experience handling multi-gigabyte unparsed vehicle logs (MDF4, PCAP, or raw CAN logs) and running offline signal processing/feature extraction algorithms. Education: Bachelor’s or Master’s degree in Computer Science, Data Engineering, Software Engineering, or a highly related technical field. Preferred Attributes Direct experience migrating or balancing hybrid architectures (seamlessly shifting data between on-premises clusters and AWS/Azure/GCP environments). Familiarity with building interactive, data-heavy frontends (using React, Vue, or Python Streamlit) to let non-technical stakeholders run complex ML simulations. Background working in manufacturing, automotive R&D labs, or hardware-in-the-loop (HIL) testing environments.