The Foundation team operates at the lowest levels of the platform stack, providing hardware and software platform efficiency and engagement. This is done both internally at Facebook and externally with the larger tech community through open source and open compute initiatives. The entire Facebook platform is the Foundation team's lab for research, development, and innovation and their work spans across connectivity at all levels. The more Facebook grows, the more costs increase, and infrastructure is the largest variable. To combat this challenge, the Foundation team is constantly innovating and building solutions to keep costs down, while supporting traffic growth and network complexity, which in turn creates a positive user experience.

Engineers and technical leaders on the team are given a great deal of autonomy in determining what they think will be impactful and therefore, what they build. This bottom-up culture gives engineers the opportunity to truly own projects and challenge themselves to find new ways to solve complex problems. As a result, engineers and technical leads come to work everyday feeling empowered to build.

The best way to understand the Foundation team and how they operate is to get to know a few of the individual teams that operate under the larger Foundation team umbrella.

The Hardware Design team’s main priority is to build and share the world's most efficient, serviceable, and scalable server platform used by Facebook’s data centers worldwide. Hardware engineers at Facebook design, develop, build and test components, automated software systems and products based on a broad range of technologies. The team innovates with world class precision, while delivering cost-effective, serviceable, reliable, and scalable hardware solutions. Many of their hardware designs are open sourced and released under the Open Compute Foundation, and they are heavily involved in other communities such as Telecom Infrastructure Project (TIP), OpenBMC, and the Linux Foundation.

The Sourcing and Operations Engineering (SOE) team leads all engineering responsibilities related to building and operating Facebook's infrastructure. A central part of the team's infrastructure growth has evolved from standard, off-the-shelf infrastructure to custom-designed, manufactured, and supported infrastructure, which is then shared with the data center community through the Open Compute Project. Considering that just one building at a Facebook data center is the size of four football fields, this is no small feat.

Facebook's Capacity Engineering & Analysis (CEA) team bridges the digital world (all software products at Facebook) with the physical world (servers, data centers, power, and networking). On the software side, the CEA team enhances performance for different products and services, improves architecture, and optimizes the interaction with other products. On the hardware side, the team determines the capacity expansion plan—including servers, data centers, and networking—to scale Facebook infrastructure.

The Infrastructure Strategy (Infra Strat) team at Facebook includes data scientists, data engineers, and software engineers, who develop data-driven solutions for significant infrastructure challenges. The Infra Strat team works closely with new areas of infrastructure and creates core assets that are hugely impactful to the company at-large, like app and site performance and the maintenance of critical large-scale data ecosystems.

Technical program managers (TPMs) are responsible for seeing programs that are responsible for building Facebook's data centers, building WhatsApp's new payment service or building a new ASIC program from beginning to end and ensuring a highly efficient workflow and effective communication. Being a TPM at Facebook means driving impact and delivering measurable results across a wide range of areas. From impacting product decisions to infrastructure deployment, to product releases and performance, TPMs strive for flawless execution against company-wide goals.

The Tech Strategy team is divided into three groups responsible for three different functions. The first team is focused on optimizing computing for Facebook's ML and AI applications. This includes working closely with Ads, Feed, Search, Instagram and a number of other teams at Facebook to optimize and determine bottlenecks within Facebook's Deep Learning algorithms. The second team falls under the ASIC engineering department. This team is focused on building Facebook's first customized silicon for our data centers. They look for ASIC development experts that span the entire product development lifecycle—from concept to GDSII and Post-Silicon Validation. Finally, the third team is made up of compute, storage and network technologists who are responsible for identifying internal infrastructure needs and assessing available and future industry technologies to meet the growing infra demands of the company. They collaborate with many teams including software, hardware, and supply chain to build compelling solutions.

AI Infra Foundation operates in a highly cross-functional manner with multiple teams including the rest of AI Infrastructure, CEA, Tech Strategy (SHARE), Hardware Engineering, and Product Infrastructure. Their main activities include workload characterization, benchmarking, optimization, simulation, performance analysis tools, telemetry, and hardware exploration. The team's efficiency efforts span all layers of the Infrastructure stack, with the goal of identifying and resolving bottlenecks in compute, memory, storage, and networking, all while ensuring overall fleet-wide system efficiency.