Loading…
Latest reads
Every post, summarised
Grounded summaries of engineering blogs, newest first. Each links to the original.
- 4,671 posts
- 20 companies
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
Latest posts
Introducing Real World VoiceEQ: Measuring the human quality of voice AI
Existing voice-AI benchmarks increasingly suggest near-human performance, while real-world conversations still expose failures involving speaker consistency, hesitation, accents, noise, emotion, and natural responses. Real World VoiceEQ measures human quality in voice interaction by testing whether systems recognize, produce, and respond to acoustic information omitted by transcripts, including tone, emotion, speaker identity, and background context. It evaluates more than 40 proprietary and open-source models across 15+ dimensions and 60+ metrics covering ASR, TTS, S2S, and Speech Understanding, using more than 1 million human ratings, including 785,000 TTS and 48,000 STS ratings. Findings indicate that capabilities are specialized, speech-to-speech systems often underuse paralinguistic cues, and traditional benchmarks can overestimate performance in noise and other conversational conditions. The research also reports that human agreement remains essential for subjective acoustic and social judgments because SLM evaluators were less reliable outside clear, verifiable tasks.
How Bullet uses ClickHouse Cloud to give DeFi's fastest exchange real-time analytics
Bullet operates a DeFi exchange for perpetuals, spot trading, and lending, with trades settled on a blockchain while users retain custody of their assets. Its sub-millisecond trading system produces more than 150 million event rows per hour, creating a need for historical indexing, real-time analytics, and operational monitoring beyond the blockchain’s state-machine model. ClickHouse Cloud consumes events from Kafka, stores them in dedicated tables, and supports user features, leaderboard and reward calculations, P&L charts, audit logs, and Grafana-backed alerts. Replacing Databricks reduced query latency from 10–15 seconds to milliseconds, improved data freshness from 1–2 hours to under 5 seconds, and removed a separate DynamoDB serving layer at comparable cost while handling 1,000 times more data across staging, testnet, and mainnet.
Take insights anywhere with Genie One on mobile
Genie One mobile apps for iOS and Android let business users ask questions of company data, view dashboards, access Databricks Apps, and use conversational agent capabilities away from a desk. Answers draw on Genie Ontology, Genie Agents, enterprise governance, and connectors to Google Drive, Microsoft 365, and Atlassian, while respecting source permissions. AI/BI Dashboards reflow widgets into a single column in portrait mode and preserve their designed layout in landscape. Authentication uses the existing OAuth flow and identity provider, with MFA, conditional access, device posture, network controls, regional data handling, and workspace permissions carried over from the browser; there is no separate mobile backend or mobile-only endpoint. The app is available in Public Preview, with dark mode, push notifications, voice mode, and account-level access listed as planned capabilities.
@clickhouse/rowbinary: when your library is also a parser compiler
ClickHouse's @clickhouse/rowbinary is a Node.js reader and writer for the RowBinary, RowBinaryWithNames, and RowBinaryWithNamesAndTypes formats, offering both a generic parser and an accompanying Agent Skill. RowBinary avoids per-row overhead but is difficult to consume in JavaScript because nested types, LEB128 varints, precision-sensitive DateTime64 values, and recursive self-describing types require extensive runtime dispatch; JSON alternatives can also round large UInt64 values. The package provides small, composable, monomorphizable reading primitives, while SKILL.md instructs a coding agent to assemble query-specific parsers with inlined operations and selected policies such as BigInt handling and array materialization. Generated parsers run 1.5–3.4x faster than composing library functions in loops, with a fixed-width benchmark parser reaching 3.41x faster, and produce reviewable source that can be tested and benchmarked like ordinary application code.
Lessons From the Leaderboard: What 5,000+ Kagglers Taught Us About Improving AI Reasoning
The NVIDIA Nemotron Model Reasoning Challenge on Kaggle tasked over 5,000 participants with improving reasoning accuracy on Nemotron-3-Nano-30B using restricted LoRA adapters on Google Cloud G4 VMs. Top-performing competitors approached reasoning as an end-to-end engineering workflow by auditing synthetic chain-of-thought traces with programmatic solvers before training. Successful teams also compressed reasoning representations into compact signatures and bit-manipulation formats to prevent long chains from exceeding runtime token generation limits. Additionally, leading approaches separated precomputed reusable structures from live computation and conducted granular per-category validation to monitor regression and non-determinism. These results show that reasoning improvements stem from verifiable training data, compact context encoding, and structured validation rather than unconstrained data scaling.
From weeks to a day: how we made LLM evaluation fast enough to iterate on
Airbnb built a four-layer LLM infrastructure framework to reduce iteration and evaluation turnaround from weeks to a single day. By using per-sample caching for generated references and judge scores, they established a deterministic evaluation foundation that separates genuine model drift from measurement noise. This setup enables rapid, bounded model hotfixes using micro LoRA adapters and comprehensive end-to-end validation across component boundaries.
Shipyard: How We Built Slack’s Next-Generation EC2 Platform
Slack developed Shipyard to modernize its EC2 infrastructure by replacing continuously updated, mutable instances with immutable deployable artifacts. The platform adopts a layered image architecture starting with a standardized base image called slack-zero, which is built using AWS Image Builder and consumed by downstream service-specific machine images. Rather than running periodic Chef jobs in the background, configuration management is restricted to image baking and initial instance provisioning phases. Fleet management is handled through the Peekaboo inventory system, metric-driven progressive rollouts via Gondola, and automated instance rotation managed by a lifecycle component known as the Reaper. Developers validate cookbook updates prior to merging using Ship Quick, which provisions real infrastructure across dedicated worker fleets to verify provisioning workflows.
How to Run an Autoresearch Workflow with RL Agent Skills and NVIDIA NeMo
Setting up reinforcement learning workflows and running repetitive experimentation often consumes significant engineering time before meaningful metrics emerge. To address this, an autonomous autoresearch workflow uses a frontier coding agent running Codex with GPT 5.5 to manage NVIDIA NeMo RL and NeMo Gym on an NVIDIA Brev GPU instance. The agent executes full-stack setup, creates a novel NeMo Gym visual counting environment, tunes Qwen3-VL-2B-Instruct accuracy from 25.0% to 96.9%, and implements an off-policy reinforcement learning algorithm directly from a research paper. Reusable skills, including brev-etiquette, session-memory, and autoresearch, enforce local conventions and prevent context drift during long executions. This setup offloads environment configuration, dependency resolution, and experiment tracking while preserving human oversight for goals and strategy.
Post-Train NVIDIA Cosmos 3 in One Day Using Agent Skills
Adapting vision reasoning models for specialized physical AI video tasks often demands days of manual container configuration, training script authoring, and hyperparameter tuning. NVIDIA Cosmos 3 Nano, an omnimodal mixture-of-transformers model, pairs an autoregressive Reasoner tower with an iterative diffusion Generator tower for video question answering. By integrating NVIDIA TAO agent skills with a coding agent like Codex, developers can automate data patching, evaluation, and Low-Rank Adaptation (LoRA) post-training. In experiments using the Woven Traffic Safety dataset, LoRA fine-tuning boosted zero-shot baseline accuracy from 54.41% to 87.14% while using approximately seven times fewer GPU hours than full-parameter supervised fine-tuning. Combining this agent workflow with TAO AutoML further elevated model accuracy to 93.35% before automated deployment via NVIDIA NIM.
Meet the first cohort of recipients shaping the future with the Uber Transit Innovation Fund
The Uber Transit Innovation Fund will provide up to $50,000 each to 20 US public transit agencies, municipalities, and regional authorities to pilot on-demand mobility programs, with $1 million available overall. Its first announced cohort includes agencies and authorities from San Antonio, Volusia County, Polk County, Belton, Monterey, South Bend, the Quad Cities, and San Juan. Proposed projects cover late-night overflow rides, microtransit and paratransit supplementation, airport connections, curb-to-curb service, modernization of taxi vouchers, subsidized rides for opioid-use-disorder treatment, food-access mobility, and first-mile/last-mile trips to 16 rail stations. Applications remain open through June 30, 2026, with proposals reviewed on a rolling basis until funds are fully allocated.
How Retail Finance teams are using Agentic AI to protect omni-channel margins
Omni-channel retail has spread margin, cash, and markdown decisions across more channels, fulfillment paths, and return routes, while agentic systems increase the speed and complexity of change. The post presents ontology as a way to preserve the meaning and business context behind finance figures, keeping definitions, channels, and cost drivers current. Databricks Genie is described as a data-smart AI coworker that answers finance questions in plain language, grounding responses in an evolving ontology, source traces, permissions, and governed AI costs. It focuses on margin after fulfillment and returns, inventory cash tied up in the wrong place, and full-price revenue at risk from markdowns and returns, then prepares actions for a person to approve. Unilever deployed Genie to more than 1,200 finance and business users; analysis that took days now takes minutes, with expected multi-million-euro annual cost avoidance.
How we configure huge pages in ClickHouse Managed Postgres
ClickHouse Managed Postgres configures Linux huge pages to reduce the per-connection page-table cost and translation overhead of a large PostgreSQL shared_buffers cache. The configuration reserves a quarter of machine memory before PostgreSQL starts, verifies HugePages_Total, sets huge_pages = 'on' to prevent silent fallback, and sizes shared_buffers against the complete shared-memory segment, including buffer descriptors, WAL buffers, and lock tables. On an r7i.4xlarge with 32GB shared_buffers, 4KB pages added 31.1MB of page tables per connection and reached 6.12GB at 200 connections, while 2MB huge pages reached 111MB. A select-only pgbench test at 100 clients also rose from 373,083 TPS to 418,087 TPS, a 12% increase, while startup failure without a sufficient pool made misconfiguration explicit.
Doing the right thing when things go wrong
Fin’s incident-management process is designed to reduce customer impact quickly when deployments, alarms, or customer reports indicate a disruption. It combines customer and product signals, including heartbeat metrics for workflows such as Fin’s response generation, with engineer monitoring of rollouts; a rollback can land in a little under two minutes. Responses run through incident.io and Slack, with defined engineering lead, incident commander, business lead, and resolution owner roles, plus the out-of-hours STRIKE responder group. The lifecycle covers triage, investigation, mitigation, and learning, with a stated preference for rolling back to a known state before fixing forward and with explicit alternatives when rollback is unavailable. After recovery, bounded follow-ups and incident reviews aim to improve recurrence prevention, blast-radius reduction, detection, and response, while early escalation and clear communication make the process repeatable rather than dependent on heroics.
Foundational context: Cross-industry & function-specific accelerators for Lakebase
Databricks presents Lakebase as a fully managed, serverless, standard Postgres database for combining operational and analytical workloads on its Data + AI Platform. The platform separates compute from storage, integrates with the lakehouse through Synced Tables and Lakebase CDF, and uses Unity Catalog for governance; copy-on-write branching and autoscaling to zero are described as core infrastructure primitives. The post showcases partner-built, ready-to-deploy accelerators spanning technology, finance, marketing, sales, supply chain, human resources, customer service, and operations. Examples include PostgreSQL migration assessment, multi-agent Genie orchestration, stateful enterprise agents, autonomous data reliability, governed contact-center intelligence, and project operations management. These offerings package Lakebase patterns into migration controls, domain-specific solutions, and agent frameworks intended to accelerate modernization and reduce transformation complexity.
Blocking Slow-Burn Attacks: Contextual Policies in Omnigent
Omnigent’s post examines how a vendor-review assistant can leak confidential pricing terms when an attacker hides an indirect prompt injection in a shared runbook. Because the malicious workflow is divided into ordinary actions, stateless checks approve each step even though the session as a whole is unsafe. The demonstration compares an unprotected run, which sends the summary externally, with a contextual policy that stores a running risk score, adds 30 for each document read, and denies email after the score exceeds 50. It also shows that agents cannot remove or disable policies, new policies require human approval, and any denial prevails when policies are combined. Runtime enforcement therefore preserves the block even when the agent has been misled.
Systems Engineering Playbook: Optimizing Qwen 3.5-397B MoE on Ironwood (TPU7x)
Serving Qwen3.5-397B, a 397-billion-parameter sparse Mixture-of-Experts model, on Ironwood (TPU v7x) requires fitting roughly 400 GB of weights into HBM while sustaining utilization across contrasting workloads. The report applies a modular, model-agnostic playbook built from reusable JAX/Pallas kernels and hardware-aware cost models, focusing new work on Gated DeltaNet linear attention and Attention Data Parallelism while integrating changes into vLLM and SGLang. The architecture activates exactly 17 billion parameters per token and combines 15 repeating blocks of three GDN layers and one GQA layer. Benchmarks use a single host with four Ironwood chips, eight logical devices, 8K/1K prefill-heavy and 1K/8K decode-heavy workloads, and concurrency tiers from 64 to 512. Between April and June 2026, stated optimizations improved inference performance by approximately 4.7x for 512-concurrency prefill-heavy workloads and 3.1x for decode-heavy workloads.
Inside Shopify Hack Days: Building a prototype for music-playing pages
During Shopify Hack Days 39, a 13-person team addressed why artists selling music on Shopify product pages lacked a built-in way to hear digital albums before buying. The team pursued two paths in parallel: a first-class Audio media type across Core, Admin API, admin-web, and storefront, and a standalone Shop Sounds app for release management. The app orchestrated product and variant creation, audio and artwork metafields, storefront access, and Online Store publication, while a theme extension provided track-level purchasing and a shared audio instance that continued across navigation. It also included six GLSL fragment shaders driven by bass, mid-range, treble, and overall energy, plus synced lyrics and a geolocated tour dates block. After three days, the team had a working prototype, but Shop Sounds was not a live product; the project demonstrated that native audio support and Shopify metaobjects and metafields could support a music toolkit without custom infrastructure.
Training Orchestrator: Unifying Model Training at Yelp
Yelp developed Training Orchestrator to standardize machine learning pipeline execution and eliminate divergent training scripts across its applied engineering teams. Previously, monolithic Spark scripts coupled training logic directly to cluster runtimes, preventing local testing, slowing debug cycles, and causing inconsistent configuration drift across projects. The new framework introduces a declarative configuration layer backed by Pydantic models, enabling static type checking and step schema validation before runtime execution starts. Training workflows are defined as modular step hierarchies that the orchestration engine compiles into directed acyclic graphs for execution in topological order. By injecting Spark and MLflow contexts into steps, the architecture ensures reproducible tracking, automated Slack alerts, and parity across local development and production environments.
Building Service Topology at Scale: Architecture, Challenges, and Lessons Learned
Engineers at Netflix required a unified, real-time view of service dependencies to navigate distributed architecture and improve incident troubleshooting. Traditional batch systems introduced stale data, so the team created a streaming-first platform backed by reactive streams and backpressure handling to ingest flow records from multi-region Kafka streams and Server-Sent Events without data loss. The architecture partitions data into physically separate graph and columnar storage layers covering eBPF network flows, IPC metrics, and distributed traces. Network flow ingestion relies on a three-stage distributed aggregation pipeline using consistent hashing to resolve network intermediaries into logical application connections. The resulting production system serves time-travel and topology queries with sub-second latency while continuously updating dependency views.
NVIDIA Ising Decoding Cuts Color Code Logical Error Rates by Over 300x
Fault-tolerant quantum computing requires effective quantum error correction, but color codes have historically been constrained by difficult decoding demands. NVIDIA introduced the Ising Decoder ColorCode 1 Fast, a pre-decoder designed to accelerate and improve color code logical error rates. The framework uses a 3D Convolutional Neural Network trained via cuStabilizer in NVIDIA cuQuantum alongside PyTorch to handle localized error syndromes. In benchmark testing at code distance d=31 with a 0.3% physical error rate, the decoder achieved over 347.7x lower logical error rates and 7.3x faster runtime compared to the Chromobius decoder. These local, space-time predictions scale to arbitrary distances, supporting parallel blockwise decoding for real-time quantum algorithm execution.
Top topics
- Architecture834
- Performance816
- LLMs660
- Machine Learning622
- Open Source563
- AI Agents553
- AI470
- Developer Experience448
Weekly digest
The best engineering reads, once a week.