Loading…
Latest reads
The engineering internet, summarised so you can actually read it.
Streamlining Resource Binding with End-to-End Support for Vulkan Descriptor Heaps
Vulkan introduces the VK_EXT_descriptor_heap extension to refactor GPU resource binding, replacing set-based layouts and pool allocations with application-managed descriptor memory heaps. This approach aligns Vulkan closer to Direct3D 12 and benefits workloads like dynamic texture indexing and complex ray tracing shaders. Developers manage memory directly, binding a single resource heap and a separate sampler heap globally to avoid high rebinding overhead. Shaders can access descriptors either through mapping structures using constant offsets or push constants, or directly via untyped pointers enabled by VK_KHR_shader_untyped_pointers. Support is integrated into NVIDIA drivers 610, Vulkan Headers 1.4.340, and Nsight Graphics 2026.2 for frame capture and inspection.
Tanya LenzQ&A: How KRAFTON Built PUBG Ally, a Co-Playable Character Powered by NVIDIA ACE
KRAFTON developed PUBG Ally, an on-device AI co-playable character for PUBG: BATTLEGROUNDS, using NVIDIA ACE tooling and models. Conventional non-playable characters rely on fixed dialogue, but PUBG Ally processes live game states and player voice input in real time. The interactive pipeline combines automatic speech recognition, a quantized Mistral-NeMo-Minitron-2B small language model, and a custom text-to-speech model running locally on client GPUs with at least 8GB of VRAM. To minimize latency, the architecture separates reactive gameplay into a behavior-tree System 1 layer while routing strategic decisions and speech generation to the language model. KRAFTON tuned the non-deterministic system across multi-layer evaluations and large-scale playtests with over a thousand players.
Elizabeth GoodmanDropbox ·
How we used DSPy to turn AI evaluations into better responses in Dash chat
Dropbox improved its Dash chat agent by establishing an automated optimization loop powered by DSPy and LLM-as-judge evaluations. Engineers first calibrated their LLM judges against human-annotated interaction traces, then used those judges to systematically optimize the agent's system prompts via offline counterfactual replay. This automated workflow reduced incomplete responses by 26% while decreasing overall token consumption.
Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy MeyerzonAchieving Near-Linear Training Scalability for Pinterest’s Foundation Models
Pinterest’s foundation ranking models store approximately 99% of their parameters in embedding tables, causing cross-node NCCL all-to-all communication to severely bottleneck multi-node distributed training. Initial multi-node scaling stalled at 1.21x on four nodes even after adopting AWS Elastic Fabric Adapter for OS-bypass networking. To resolve this, engineers profiled GPU activity with PyTorch Profiler and systematically addressed communication bottlenecks across multiple layers. The team implemented FP8 quantized communications using FBGEMM, balanced table sharding across GPUs, halved embedding dimensions while doubling row counts, and adopted a 2D parallel communication topology alongside PyTorch Distributed Checkpoint. These combined interventions achieved near-linear scaling factors of 3.9x on four nodes and 7.5x on eight nodes across 64 GPUs.
Pinterest EngineeringAutomated Schema Evolution in Pinterest’s Next-Generation DB Ingestion Framework
Pinterest's CDC ingestion platform relies on a multi-stage pipeline built with Kafka, Flink, Spark, and Iceberg, where upstream schema updates risk disrupting data transformations and offline storage. To prevent pipeline failures and schema drift, Pinterest introduced an automated schema evolution framework based on a staged convergence model. Upstream changes are detected via push-based DDL messages and daily pull-based validation checks. The workflow automatically restricts automated updates to additive changes and numeric precision widening, updating Iceberg table definitions and regenerating Flink and Spark code through auditable pull requests. By decoupling schema divergence, code deployment, and data convergence, the system maintains continuous ingestion availability while working toward eventual schema consistency across all storage layers.
Pinterest EngineeringAccelerating BEV Pooling on NVIDIA GPUs for Physical AI Applications
Bird's-eye-view (BEV) pooling frequently becomes an inference latency bottleneck in spatial AI pipelines due to irregular memory access patterns and scatter-reduce operations. BEVPoolV3 addresses this by tailoring kernel execution to the GPU memory regime, specifically distinguishing between DRAM-bound and L2-cache-resident working sets. The implementation reduces redundant scatter traffic via depth-outer traversal, precomputes indices to eliminate runtime integer division, minimizes duplicate depth loads, and adopts interval-owned writes to bypass atomic operations. On a canonical 49 MB nuScenes working set, BEVPoolV3 accelerates TensorRT plugin execution from 274.0 µs down to 16.4 µs on an NVIDIA RTX PRO 6000 Blackwell Max-Q GPU. Furthermore, profiling with Nsight Compute reveals that L2-resident scatter-reduce operators achieve optimal efficiency with FP8, avoiding the ALU instruction overhead of NVFP4 decoding.
John YangMaximize AI Factory Energy Efficiency Through Full-Stack Inference and Training Optimizations
Power costs comprise up to 40% of AI factory operating expenses, making performance per watt critical within fixed site power limits. Operators maximize throughput by combining architectural enhancements, narrow-precision formats like NVFP4, and mixture-of-experts model designs that activate fewer parameters per token. During distributed LLM training in Megatron-LM, tuning individual GPU speeds along critical execution paths eliminates idle power waste without extending total training time. Additionally, the NVIDIA DSX platform orchestrates dynamic power allocation, 45°C direct-to-chip liquid cooling, and grid integration across racks and facilities. Together, these full-stack optimizations enable performance-optimized AI factories to deliver up to 2.6 times more tokens per second per megawatt.
Sachin IdgunjiBoost Inference Performance up to 15x on NVIDIA Blackwell Using DFlash Speculative Decoding
Sequential token generation in autoregressive LLMs limits GPU utilization and serving throughput, while conventional speculative decoding remains constrained by autoregressive draft models. DFlash addresses this bottleneck by employing a lightweight block-diffusion drafter that predicts an entire block of masked candidate tokens in a single forward pass. The architecture incorporates target hidden-state conditioning and key-value injection into draft model projections while relying on the target model for parallel verification. Evaluated on NVIDIA Blackwell GPUs across TensorRT-LLM, vLLM, and SGLang, DFlash increases gpt-oss-120b throughput by more than 15x at high interactivity targets and delivers up to 5.8x speedups on Gemma 4 31B compared to autoregressive decoding.
Amr ElmeleegyYour AI shipped a backend that boots. That is the whole problem.
Backend security vulnerabilities often emerge from permissive default settings in web frameworks, a risk amplified when AI agents generate code focused solely on passing tests. Common failure points include unmetered request body parsing, unvalidated schemas, prototype pollution, and wildcard cross-origin resource sharing policies. To eliminate these foot-guns, the TypeScript framework DaloyJS enforces secure-by-default behavior across its entire routing and middleware model. It provides built-in request timeouts, streamed body size caps, prototype-stripping JSON revivers, and unified Zod schemas that simultaneously validate payloads and generate OpenAPI contracts. While secure defaults lower the risk of common infrastructure exploits, developers still remain responsible for application-level authorization and business logic correctness.
Devlin DuldulaoThe 2026 Developer Survey is now open (for human developers only)!
The 2026 Developer Survey is open to gather data on software builders' tools, working lives, and emerging practices. Marking fifteen years since its initial 2011 release, the annual survey focuses heavily on the role of artificial intelligence and autonomous agents across the software development lifecycle. Past survey results indicated that while developer adoption of AI and agents doubled, user trust declined alongside escalating concerns regarding generated code quality. In addition to modern AI trends and financial return considerations, the questionnaire maintains traditional inquiries regarding day-to-day technologies and developer workflows. The survey is hosted on the Qualtrics platform, which requires users to pause third-party ad blockers and adheres to regional access restrictions across multiple countries.
Ryan DonovanBuild an AI Scientist for Life Science Discovery with NVIDIA BioNeMo Agent Toolkit
General-purpose AI agents often struggle with biomolecular discovery because scientific workflows require precise model selection, strict input formatting, and artifact interpretation. The NVIDIA BioNeMo Agent Toolkit bridges this gap by providing BioNeMo Skills and Model Context Protocol wrappers on top of accelerated NVIDIA NIM microservices. These skills instruct agents on tool capabilities, required inputs, expected file formats like CIF and SMILES, and specific failure modes. Developers can route requests through hosted endpoints for testing or deploy local NIM microservices on GPU nodes to lower warm per-call latency during iterative candidate generation loops. Internal benchmarks using Codex CLI demonstrate that equipping agents with BioNeMo Skills improves task completion rates from 57.1% to 100% while doubling token efficiency across passing assertions.
Kyle TretinaHow Telcos Build Autonomous Networks with Agentic AI
Telecom operators currently operate mostly within Levels 2 and 3 of the TM Forum autonomous networks taxonomy, relying on predefined scripts. Progressing to Levels 4 and 5 demands autonomous systems capable of intent comprehension, real-time telemetry sensing, and cross-domain action coordination. To address this, an autonomy platform architecture combines domain-specific reasoning models, agent harnesses, and isolated sandboxed runtimes. Workflows separate responsibilities among on-demand agents, long-running execution loops, and deep-research multi-agent systems that evaluate ranked remediation paths or design wireless algorithms. Initial experiments in simulated SR-MPLS routing and wireless physical or MAC layer design show autonomous discovery can deliver measurable gains over standard operational and engineering baselines.
Amogh DendukuriNetflix ·
Toward More Controllable AI Video Editing: An Early Research Exploration at Netflix
Generative video editing tools often regenerate entire video clips, which inadvertently modifies untouched scene details or violates physical continuity during object deletion. To provide artists with precise control, two research prototypes were developed: Vera and VOID. Vera uses a layered video diffusion framework with a Mixture-of-Transformers architecture to jointly generate separate edit and alpha matte layers, compositing them with original footage while leaving unchanged pixels untouched. VOID addresses object deletion by conditioning on interaction-aware quadmasks and using a second-pass appearance refiner to reconstruct scenes with plausible physical continuity. In a user study evaluated across 75 real-world scenarios, reviewers selected VOID 64.8% of the time over six baselines.
Netflix Technology BlogNetflix ·
How Netflix Simplified Batch Compute with Kueue
Netflix transitioned its managed batch compute infrastructure from a homegrown solution called Compute Managed Batch to Kueue on its Titus container platform. CMB previously relied on custom scheduling and admission-only fair sharing without preemption, making feature development cumbersome as the Kubernetes ecosystem evolved. To modernize the platform transparently, Netflix mapped internal tenants to Cohorts and leaf tenants to ClusterQueues and LocalQueues while routing jobs through a custom Kueue router. Kueue operates alongside existing Titus scheduling profiles rather than replacing the kube-scheduler, preserving cluster placement efficiency. The migration was completed in four weeks across millions of batch workloads, significantly increasing average resource utilization through preemption-based fair sharing.
Netflix Technology BlogCCCL Runtime: A Modern C++ Runtime for CUDA
NVIDIA CCCL runtime introduces modern C++ abstractions for fundamental CUDA programming concepts starting in CUDA Toolkit 13.2 and CCCL 3.2. Designed as an alternative to the traditional CUDA runtime API, the library eliminates implicit global state by requiring explicit dependencies, such as binding streams directly to specific device references. The API enforces strong typing through owning types and non-owning reference types, omits default stream usage to ensure all streams remain non-blocking, and handles errors via standard C++ exceptions. In addition, CCCL runtime simplifies kernel launches through kernel functors that enable automatic template argument deduction and automatically transforms owning buffer arguments into device spans.
Piotr CiolkoszEnable Real-Time AI for High-Speed Data Acquisition with DAQIRI
Scientific instruments and high-bandwidth sensors generate immense data volumes, but traditional architectures rely on a store-then-analyze model that bottlenecks real-time discovery and forces extensive data discarding. NVIDIA DAQIRI shifts data acquisition to a software-centric pipeline that connects streaming detectors directly to GPU computing. Built as part of the NVIDIA Holoscan Platform, DAQIRI uses DPDK to bypass the Linux kernel and move UDP and RoCE v2 traffic directly from the NIC into GPU DMA buffers at line rates exceeding hundreds of gigabits per second. The system automates packet reordering, flow steering, and on-GPU data conversions such as mapping wire int4 payloads into fp16 tensors using YAML configuration files and simple C++ or Python APIs. This enables real-time edge processing, filtering, and model inference across experimental pipelines like CERN ATLAS.
Cara LaaschInside NVIDIA Halos for Robotics: A Full-Stack Functional Safety System for Physical AI
NVIDIA announced the launch of NVIDIA Halos for Robotics, a full-stack functional safety system designed for industrial robots, humanoids, and autonomous mobile robots. The platform extends NVIDIA's functional safety architecture from autonomous vehicles into robotics, utilizing NVIDIA IGX Thor and the NVIDIA Holoscan Sensor Bridge as its hardware foundation. On top of this hardware, Halos OS provides the base Halos Core safety operating system and application blueprints, including an Outside-In Safety Blueprint that leverages external worksite cameras and safety decision modules. Platform hardware safety features include an isolated IEC 61508 SIL 3 capable Safety Island alongside extensive fault-detection mechanisms. To accelerate compliance, the accredited NVIDIA Halos AI Systems Inspection Lab assesses partner integrations before final third-party certification.
Suhas Hariharapura SheshadriGrab ·
Scaling out Distroless adoption With AI
Grab is transitioning its microservices to Distroless base images to eliminate unnecessary binaries and reduce vulnerability risks, but the migration risks runtime failures from missing shared objects and system utilities. To safely validate container execution in continuous integration without staging dependencies, the team relied on medium tests that run containerized services alongside internal dependencies managed by Testcontainers. Because hundreds of services lacked this test harness, Grab implemented an agentic workflow using Claude Code and Model Context Protocol integrations to inspect repositories, generate test boilerplate, and resolve configuration errors. Once test baselines are established, an automated patch-test-compare pipeline updates Dockerfiles, constructs multi-stage builds for necessary dynamic libraries, and creates draft merge requests for human approval.
Jia Yee ChongGrab ·
Palana (Part 2): Architecting isolation, identity, and auditability for AI agents
Grab's Palana platform provisions isolated, Kubernetes-native runtime environments for autonomous AI agents using dedicated per-agent namespaces and role-based access controls. The architecture separates network enforcement across layers, applying Layer 3 and Layer 4 containment with Cilium and NetworkPolicy alongside Layer 7 application filtering evaluated by Open Policy Agent. Agent interactions with large language models route through a LiteLLM proxy wrapper that retrieves credentials from HashiCorp Vault based on Kubernetes pod context rather than client headers. Secrets management is divided between directly readable agent paths and proxy-only placeholder paths that prevent raw tokens from residing in runtime filesystems. An automated reaper monitors multi-source activity signals to shut down idle compute resources while preserving persistent storage and configuration state.
Kevin LittlejohnNetflix ·
The Data Canary: How Netflix Validates Catalog Metadata
A manual mitigation action during an incident corrupted a data feed for a subset of titles, causing playback issues and catalog service failures that existing code canary systems failed to catch. To protect streaming reliability, Netflix built an automated data canary system that validates transformed catalog metadata prior to publication. The architecture utilizes a dedicated orchestrator alongside permanent baseline and canary service clusters to coordinate validation using real production traffic. By leveraging custom chaos experiment thresholds, sticky session affinity, and Starts Per Second playback metrics, the system detects regressions in under ten minutes and blocks publication automatically. Controlled failure injection experiments routing approximately 0.2% of global traffic confirmed that issues could be identified in 2.5 to 4 minutes.
Netflix Technology Blog