@Scale: Technical Conferences for Engineers Building at Scale

@Scale describes itself as a series of technical conferences for engineers who build or maintain systems that are designed for scale. The site groups material by event track, including Systems @Scale, and publishes talk announcements, technical blog posts and event recaps. Articles are dated by year and attributed to the @Scale team or named engineering authors.

• Technical conference series for engineers working at scale
• Tracks include Systems @Scale
• Publishes talk announcements and technical articles
• Content archive organised by type and year

The site lists an event on 25 June 2026 at the Meydenbauer Center in Bellevue, Washington. The stated theme is building the advanced infrastructure necessary to power today's sophisticated AI models, framed as an engineering challenge requiring scalable, high-performance and reliable computing systems tailored to machine learning workloads. Topics named include custom silicon accelerators such as GPUs and TPUs, ultra-high-speed networking fabrics and storage architectures for massive data throughput. The programme description also covers the use of AI to design, operate and optimise large-scale systems, including dynamic resource allocation, predictive maintenance and anomaly detection.

@Scale has run virtual Systems @Scale events at which engineers present work followed by a live question-and-answer session. Announcement posts give the talk time — for example 11am PT — and invite questions to be submitted in advance by email to systemsatscale@fb.com. Talks announced this way include Optimizing video storage via Semantic Replication and FlightTracker: Social graph consistency at scale.

• Virtual event format with live Q&A
• Questions submitted in advance to systemsatscale@fb.com
• Talks scheduled at 11am PT

A large part of the published material covers backend infrastructure at Meta and Facebook scale. Articles describe FlightTracker, a set of APIs and services providing read-your-writes consistency across TAO and global secondary indexes; Conveyor, a continuous deployment platform handling over 100,000 deployments a week; and Owl, which addresses distribution of large, hot content such as executables, code artifacts, AI models and search indexes within Meta's private cloud. Other pieces cover AI observability infrastructure for profiling machine learning workloads and a stateful web service built as a distributed PHP lambda function on top of BladeRunner for real-time client-server interactions.

• FlightTracker: read-your-writes consistency for the social graph
• Conveyor: continuous deployment pipelines and push plans
• Owl: fast, efficient and reliable content distribution
• AI observability: telemetry, profiling and performance introspection
• Stateful web service on BladeRunner for real-time interactions

Video engineering is a recurring subject. One article describes semantic replication, a technique for reducing the number of stored copies of video encodings in Everstore, Facebook's BLOB storage system, reported as saving over 30 per cent of video storage. Another covers the adoption of the AV1 codec, standardised by the Alliance for Open Media in 2018, for real-time communications, including screen-content coding tools, reference picture resampling and challenges around binary size, CPU and memory usage.

• Semantic replication for video storage in Everstore
• AV1 for real-time communications and BD-BR bitrate savings
• Scalability layers: temporal, spatial and quality

The site also publishes articles on client-side engineering. MobileConfig is described as a cross-platform configuration-management system, in production since 2015, that distributes strongly typed configuration values to mobile clients and integrates with feature flagging and experimentation platforms. Other posts cover msys, a shared messaging layer begun by Messenger engineers in 2018; Design Systems @ Scale, on supporting more than 20 design systems through a Design Systems Platform, theming infrastructure and chameleon components; and an account of reducing Instagram iOS build times.

• MobileConfig: cross-platform configuration management
• msys: shared foundational messaging layer
• Design Systems Platform, theming and chameleon components
• Improving Instagram iOS build speeds