Open Nav is the go-to MLOps solution in newly announced autonomous driving ecosystem is the go-to -MLOps solution1

Introducing a new CI/CD toolchain for Autonomous Driving with MLOps

Autonomous driving is a leading deep learning and machine learning use case that requires large amounts of data, diverse computational resources and most importantly, a toolchain that fluidly supports CI/CD capabilities. In cadence to the demand of the automotive industry for ADAS features, Dell Technologies has released a new reference architecture to support the specific needs of automotive OEMs, Tier 1 suppliers, and autonomous driving developers. As part of the reference architecture, has been selected to deliver a Machine Learning Operations (MLOps) solution for heterogeneous infrastructure (GPUs, CPUs, AI accelerators) for ADAS and AD developers. The toolchain has been thoughtfully created by Florian Bauman, CTO of Automotive and AI at Dell Technologies to consider the end-to-end steps of a typical ADAS workflow.

Key Challenges of Autonomous Driving Architecture

In order to build safe and reliable ADAS features, automakers need to gather and manage massive amounts of data, run diverse workloads in parallel, and train and deploy AI applications that can continuously learn in real time. With the ideal tools for these challenges typically available in silos and not well connected to a flow, Dell finds that ADAS developers are often halted by an incomplete toolchain. There is no real end-to-end workflow that ties the tools in a cohesive pipeline and simplifies developer productivity. According to a report on MLOps, only 7% of enterprises are able to successfully scale AI. This is largely due to ML operational, collaborative, and infrastructure complexities. MLOps platforms are key to operationalizing the models to production.

Autonomous Driving ML development is an iterative process that requires a toolchain that supports continuous integration, continuous delivery, and continuous deployment in real time. There is a jungle of disconnected tools built to manage the machine learning workflow, but often these tools create more complexity and technical debt. This can cause endless bottlenecks and a siloed hot-spots, without delivering a truly end-to-end solution. As a result, nearly 80% of models never make it to production. It’s critical for automakers to look at the entire infrastructure to ensure it integrates seamlessly and supports the complete pipeline from data to production.

 Introducing a Continuous Development Toolchain for Autonomous Driving

dell autonomous vehicles

The Dell Autonomous Drive Ecosystem is a reference architecture constructed of leading industry solutions that can be integrated to provide an ideal toolchain for a faster, more efficient ADAS workflow. In conjunction with leading industry and technology partners, Dell’s architecture combines Dell Technologies and partner infrastructure, software and services for a complete end-to-end ML/AI development and deployment toolchain.

As part of the Dell Autonomous Drive Partner Ecosystem, Dell Technologies selected as the MLOps solution deployed across GPU servers, CPU servers, storage and AI accelerator clusters, within the autonomous driving ecosystem. is the mission control overseeing the workflow and makes development easier for data scientists who otherwise spend 80% of their time on DevOps, non-productive tasks, supporting the toolchain.

Delivering MLOps in the Dell Autonomous Driving Reference Architecture

Along with other vetted technologies, delivers the unique capabilities required for ADAS development. Our focus on seamless CI/CD integration, dataset caching and hybrid computation capabilities (cloud, on-prem and edge) makes an ideal solution for developing new AD/ADAS features. With, Dell Technologies customers can reach production faster while data scientists can drastically improve productivity.

“Dell’s Autonomous Drive Ecosystem has been developed to deliver a state-of-the-art reference architecture for our customers to create top of the line AI features.” says Florian Baumann, CTO of Automotive & AI at Dell Technologies. “We’re confident in’s capabilities to deliver the MLOps solutions within our ecosystem for a complete end-to-end tool kit.”[1]  

Key requirements for an ADAS MLOps solution:

  • Supports CI/CD of ADAS ML features
  • Assures dataset proximity to compute for high performance
  • Hybrid computation capabilities (cloud, carrier, on-prem and edge)
  • Dataset caching tier for machine learning projects
  • Seamless open source software integration and management for ML development
  • Faster development for data scientist experimentation and delivery
  • Automated real time CI/CD pipelines
  • Management and optimization of diverse workloads

The selection of in the Dell Autonomous Drive Ecosystem not only demonstrates the capabilities of MLOps technology, but also its proven success as a solution for autonomous driving technology developments” says Yochay Ettun, CEO and Co-founder of “It’s an incredible achievement to be trusted by Dell to deliver state-of-the-art MLOps to their world class autonomous driving customers, and to be in an ecosystem amongst other leading industry and technology partners on the road to autonomous driving. has collaborated with Dell Technologies as a technology partner on various solutions to deliver a new ML architecture that accelerates ML/AI workloads with a Kubernetes-based dataset caching tier, version control, querying and more.

Start your MLOps journey with our AI starter-kit, which is the easy test drive for your development with either GPU or CPU servers.

Top MLOps guides and news in your inbox every month