MLOps Community
+00:00 GMT
Sign in or Join the community to continue

How Aurora Accelerates Autonomous Vehicle ML Model Development Using Kubeflow

Posted Jan 27, 2023 | Views 833
# Kubeflow
# Aurora Innovation
# Self-driving Cars
Share
speakers
avatar
Ankit Aggarwal
Senior Software Engineer @ Aurora Innovation

Ankit Aggarwal is Senior Software Engineer at Aurora. He has worked in the ML Platform space for over 6 years. Most recently, he has been leading the effort to accelerate the autonomy workflows at Aurora with Kubeflow Pipelines. His earlier work included building ML Inference as a Service platform for large scale metrics and simulation at Uber ATG and contributions to several open-source projects.

Ankit has a MS degree in CS from Stony Brook University and a BE degree from NSIT, India. Outside of work, he enjoys running, tennis, music, art and cinema.

+ Read More
avatar
Vinay Anantharaman
TLM MLOps @ Aurora Innovation

Technical Lead and manager of MLOps at Aurora. He currently develops the infrastructure for ML Training and large scale model inference. Previously he founded a computer vision startup focused on food recognition for consumers. He built a mobile applications utilizing computer vision and the backend systems for model inference.

+ Read More
avatar
Maurizio Vitale
Staff Engineer @ Aurora Innovation

Maurizio joined Aurora when it was still a relatively small company, fewer than 100 people in 2018.

He initially worked in the simulation team, helping scaling the infrastructure from a few thousand simulation runs per day to a few millions. Then he joined the Motion Planning team, working on pipelines for feature extraction and more recently the compute team working on MLOps projects, increasing the adoption of kubeflow being the main focus at the moment.

In previous lives he wore many hats, from hardware designer, to compiler-like tools, passing by log aggregation for a certain search and advertising company.

+ Read More
SUMMARY

In this talk, team Aurora discusses how they accelerated ML model development for autonomous vehicles by integrating with Kubeflow.

Team Aurora covers how the Kubeflow infrastructure evolved and how it is currently deployed. Then they discuss how they build pipelines, the developer experience, and the benefits of using pipelines. Finally, they walk through how they adopted Kubeflow org-wide.

+ Read More

Watch More

1:01:43
How to Systematically Test and Evaluate Your LLMs Apps
Posted Oct 18, 2024 | Views 14.1K
# LLMs
# Engineering best practices
# Comet ML
Building LLM Applications for Production
Posted Jun 20, 2023 | Views 10.7K
# LLM in Production
# LLMs
# Claypot AI
# Redis.io
# Gantry.io
# Predibase.com
# Humanloop.com
# Anyscale.com
# Zilliz.com
# Arize.com
# Nvidia.com
# TrueFoundry.com
# Premai.io
# Continual.ai
# Argilla.io
# Genesiscloud.com
# Rungalileo.io