Annapurna Labs is an Amazon company with development centers in the US and Israel dedicated to building innovation in silicon and hardware for our Amazon Web Services (AWS) infrastructure and customers at software speed and agility. As part of AWS, Annapurna is at the forefront of developing technology for one of the fastest growing IT organizations on the planet, combining a global presence with very talented engineers working together in a start-up culture. We are looking for exceptional candidates in all disciplines across silicon engineering, operations, verification, hardware and software development including deep-learning compiler, run-time, and tools to join us on this incredible journey.

Why join AWS-Annapurna?

We are on a mission to revolutionize the way datacenter acceleration is done in silicon and software. If you enjoy a fast-paced, innovative and start-up like environment filled with some of the brightest minds in networking, storage, compute and machine learning acceleration, then you will love working here.  If you want to have impact and contribute in significant ways, to invent and have first-hand visibility to how your implementations delight customers, and have fun while doing it, then you will fit right in. 

What are some of the projects executed by the Annapurna team?

Nitro System: Annapurna silicon is at the center of the AWS Nitro system that now offloads all AWS Elastic Compute EC2 networking, storage, security and hypervisor functions, allowing 100% of the server hardware to be devoted to our customers applications.  Annapurna’s high-performance Nitro hardware has enabled new use cases including local NVMe storage on the C5d, M5d, and bare metal EC2 instances as well as in the Xen-virtualized I3 and F1 instances. The Annapurna Nitro technology is also at the base of the EC2 bare metal instances, and enable 100Gb Networking for EC2’s C5n for HPC applications, and P3dn for ML training.

AWS Graviton Processors: Annapurna developed the new Arm-based AWS Graviton Processors that are custom built for cloud-scale applications powering the EC2 A1 Instances.  Amazon EC2 A1 instances are optimized for performance and cost. They are a great fit for scale-out workloads where you can share the load across a group of smaller instances. This includes containerized microservices, web servers, development environments, and caching fleets. Graviton: Introducing Amazon EC2 A1 Instances Based on the Arm Architecture

Machine Learning: Annapurna developed AWS’ first custom designed machine learning chip called Inferentia that supports all the mainstream deep learning frameworks including TensorFlow, Apache MXNet, and PyTorch, as well as models that use the ONNX format. Inferentia provides high throughput, low latency inference performance at very low cost. Each chip provides hundreds of TOPS (tera operations per second) of inference throughput to allow complex models to make fast predictions. For even more performance, multiple accelerators can be used together to drive thousands of TOPS of throughput. In addition to developing the silicon and hardware, the Annapurna team is also responsible for developing the deep-learning compiler, the software run-time, firmware, and all the performance modeling and tools to make it as easy as possible for our customers to integrate machine learning into their workloads. It’s much more than silicon, it’s really an end-to-end system innovation.

Elastic Network Adapter/Elastic Fabric Adapter: Annapurna develops a lot of critical and impactful software that enable new applications for AWS networking and storage. ENA is a custom network interface optimized to deliver high throughput and packet per second (PPS) performance for AWS EC2 instances. The Elastic Fabric Adapter (EFA) developed by Annapurna, is a network interface for Amazon EC2 instances that enables customers to run HPC applications requiring high levels of inter-instance communications, like computational fluid dynamics, weather modeling, and reservoir simulation, at scale on AWS. It uses a custom-built operating system bypass technique to enhance the performance of inter-instance communications, which is critical to scaling HPC applications. With EFA, HPC applications using popular HPC technologies like Message Passing Interface (MPI) can scale to thousands of CPU cores.

F1 FPGA: Annapurna developed the hardware, software, and infrastructure for the AWS EC2 F1 instances using FPGAs that deliver up to 100X acceleration compared to CPUs for a diverse set of compute-bound applications. F1 instances are easy to program and come with everything you need to develop, simulate, debug, and compile your hardware acceleration code, including an FPGA Developer AMI and supporting hardware level development on the cloud. Applications that use F1 include big data analytics, genomics, electronic design automation (EDA), image and video processing, compression, security, and search/analytics. Customers are using F1 instances when in need of processing large amounts of data at lower overall costs, as well as when the acceleration helps significantly reduce compute time.

And that’s just some of what is public.  We are working on many other innovative projects that we hope will delight our customers. Come join us on our mission to deliver cloud-scale innovation in hardware and software and be part of the fastest growing and most innovative IT organization on the planet. 


Find jobs in Annapurna Labs