Senior MLOps Engineer

As a Senior MLOps Engineer in the Perception team, you will play a pivotal role in building and maintaining the backbone of our L2+ ADAS stack. This senior role calls for an experienced engineer who can think critically, execute independently, and deliver results on scalable deep learning infrastructure, optimize massive data ingestion pipelines, and ensure maximum efficiency across our compute clusters.

You will be responsible for the entire DL infrastructure lifecycle—from managing Azure storage and hybrid Kubernetes clusters to designing efficient data loaders for multimodal training. You will work at the intersection of infrastructure, data engineering, and deep learning, enabling feature teams to train complex models (single frame, temporal, and multimodal) with speed and reliability. Your ability to solve abstract infrastructure challenges and apply "T-shaped" expertise—going deep in areas like infrastructure, multitask deep learning among others while maintaining breadth in software design—will be key to our success.

Responsibilities

Deep Learning Infrastructure & Compute:
●Manage and optimize the entire DL infrastructure, including Azure Blob Storage integration, VNET setups, and hybrid compute resources (Cloud and On-premise/Frankfurt clusters).
●Lead performance investigations and benchmarking for next-gen hardware (e.g., comparing H200 vs. H100, Azure native vs. deployment nodes) to ensure cost and speed efficiency.
●Maintain and scale Kubernetes clusters for training and inference workloads.

Data Pipelines & Efficient Loading:
●Architect and develop high-performance data loaders for complex multimodal datasets (camera, radar, temporal/non-temporal data).
●Modernize data processing pipelines using Ray and Kubernetes to parallelize data caching, shuffling, and oversampling.
●Leverage PyArrow and SQL to optimize data consumption and integration with data loops.
●Implement efficient dataset update strategies (handling deltas) and ensure seamless integration of new tasks into the multimodal multi-task network.

CI/CD, Monitoring & Quality:
●Design and maintain robust GitHub Workflows and CI pipelines for new and existing feature teams.
●Develop KPI dashboards using Grafana to monitor compute usage, GPU efficiency, unit test durations, and overall system health.
●Manage dependency updates (Torch upgrades, Ubuntu updates, Hydra maintenance, Dependabots) to ensure a secure and modern stack.
●Drive software design excellence by performing thorough code reviews (PRs) and enforcing high standards in software architecture.
Embedded & Evaluation:
●Establish scalable evaluation pipelines for embedded targets, specifically for QNN boards and other edge devices.
●Collaborate with feature teams to support model compression experiments and on-target performance verification.

Required Qualifications
●Education: Bachelor’s degree in Computer Science, Electrical Engineering, or a related field. An advanced degree is an advantage.
●Experience: 5+ years of industry experience in MLOps, Data Engineering, or Software Infrastructure, with a focus on Deep Learning systems.
●Programming & Software Design: Expert-level proficiency in Python with a strong emphasis on clean software design, object-oriented programming, and architectural patterns.
●Infrastructure & Orchestration: Deep hands-on experience with Kubernetes, Docker, and Cloud platforms (specifically Azure ML, Azure Storage/Networking).
●Big Data & Optimization: Proficiency with high-performance data processing tools such as Ray, PyArrow, and SQL. Experience optimizing data loading bottlenecks for
GPU training.
●DevOps & Monitoring: Experience setting up complex CI/CD pipelines (GitHub Actions) and observability stacks (Grafana, Prometheus).
●Soft Skills: Strong problem-solving abilities, proactiveness, and ownership of complex topics. Ability to adapt quickly to new technologies and work collaboratively in a supportive, high-performance team.

 

 

 

Bachelor’s degree in Computer Science, Electrical Engineering, or a related field. An advanced degree is an advantage.

 

• Deep Learning Knowledge: While deep algorithm knowledge is not mandatory, a strong understanding of Deep Learning workflows (PyTorch, training loops, multi-task learning) is highly beneficial.
• Embedded Systems: Experience with deploying or evaluating models on embedded hardware (Qualcomm QNN, etc.) or setting up hardware-in-the-loop pipelines.
• T-Shaped Skills: Ability to specialize deeply in infrastructure while lending a hand to feature teams on diverse topics (e.g., documentation builds, visualization tools).