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
How Stripe uses graph search and state machines to auto-remediate a global database fleet
Stripe describes how its Document Database fleet previously handled degraded MongoDB shards with single-step, hard-coded plugins that struggled with combined failures, intermediate states, and more than 40 shard layouts. The replacement models shard topology and health as a directed graph, with infrastructure operations as edges and safety invariants defining valid states. A simulator uses breadth-first search to find minimum-operation recovery paths, while an abstraction over Temporal contexts lets the same remediation logic run in memory without real infrastructure calls; later, the system also uses Dijkstra's algorithm. The system reduced pager volume by 30%—about 200 pages annually—reclaimed 12 days of fleet health each year, and handled a new layout with zero code changes, according to the post.
Migrating from Apollo Tooling to GraphQL Codegen at Yelp
Yelp migrated its frontend React monorepo from the deprecated apollo-tooling to GraphQL Codegen to ensure ongoing maintenance and build reliability. Because the monorepo spans over 500 packages with thousands of generated type files and custom nested type imports, the Client Data team prioritized a transparent migration that avoided codebase-wide changes. The team rejected approaches requiring syntax updates or altered import paths, choosing instead to configure GraphQL Codegen and contribute upstream plugin changes so output structures and naming conventions matched the legacy output. Migrating packages incrementally through small pull requests enabled continuous feedback while allowing developers across dozens of teams to maintain their workflows without editing existing code. Completing the migration to GraphQL Codegen v6 simplified Yelp's build tooling, eliminated deprecated dependencies, and left upstream compatibility enhancements available for the wider open-source community.
Build a Multi-Camera 3D Tracking Application with NVIDIA DeepStream 9.1 Skills
NVIDIA DeepStream 9.1 introduces AutoMagicCalib and Multi-View 3D Tracking skills to resolve the limitations of single-camera 2D tracking and manual camera calibration in large-scale video analytics. Multi-View 3D Tracking projects detections into a shared 3D coordinate system, matching tracklets across cameras via MQTT and maintaining globally consistent object IDs. AutoMagicCalib automates camera intrinsic and extrinsic parameter estimation by analyzing object trajectories across video feeds, optionally using a Visual Geometry Grounded Transformer. These capabilities are integrated into 13 agentic skills that allow coding agents to set up, configure, and execute end-to-end vision AI pipelines from natural language prompts. Outputs include an on-screen display grid, bird's-eye view trajectories, and per-frame protobuf metadata streamed through Kafka.
Introducing Claude Tag
Claude Tag is a Slack-based way for teams to delegate work to Claude by tagging @Claude in shared channels, with access to selected tools, data, and codebases. Claude breaks requests into stages, uses permitted tools, replies in threads, retains relevant channel context, and can pursue scheduled tasks asynchronously; ambient behavior can also surface information and follow up on unresolved work. Administrators scope Claude identities, memories, tools, and information by channel, set organization- and channel-level token-spend limits, and review activity logs, while users can also message Claude privately. Available in beta to Claude Enterprise and Team customers, Claude Tag replaces the existing Claude in Slack app, works with Opus 4.8, and is credited with creating 65% of the product team’s code through Anthropic’s internal version.
Why Trainy migrated from Amazon RDS Postgres to ClickHouse Managed Postgres
Trainy’s Pluto is an open-source experiment tracker that separates analytics metrics in ClickHouse Cloud from relational run metadata in Postgres. Its original metadata store, Amazon RDS Postgres, was too expensive for the volume observed and contributed to slower initial page loads, so Trainy migrated roughly 40–50 GB to ClickHouse Managed Postgres, which uses local NVMe storage and keeps both databases in one account. After migration, initial page loads became 2–3 times faster, while a complex query reported as taking 10–15 minutes previously returned in five seconds; visualization queries over runs with hundreds of thousands of steps typically return in 50–100 milliseconds at the P50. Trainy reports 100% reliable ingestion with zero dropped data points and says the unified stack could support future joins between GPU metrics and run metadata, as well as governed agent access.
Develop Lightweight USD Runtimes Faster with AI Agents
Building custom Universal Scene Description (USD) implementations traditionally required adapting large existing codebases to meet specific memory footprints, application binary interfaces (ABIs), or performance requirements. The nanousd-labs project introduces an alternative approach by using AI agents to generate lightweight USD runtimes directly from the Alliance for OpenUSD machine-readable USD Core Specification. Agents parse the specification section by section, write conforming code, and validate output against specification-derived tests under engineer guidance. Written in C++ with a public C API, nanousd functions as an independent data layer that handles parsing, composition, queries, and writing without rendering pixels. This methodology enables developers to regenerate runtimes for varying deployment constraints while maintaining standard compliance through reusable workflows and skill graphs.
Agents for financial services
Anthropic is releasing ten ready-to-run agent templates for financial services, targeting work such as pitchbook creation, KYC screening, and month-end close operations. Each template packages task-specific skills, governed data connectors, and subagents, and is available as a plugin for Claude Cowork or Claude Code, or as a cookbook for Claude Managed Agents. Claude add-ins for Excel, PowerPoint, Word, and, soon, Outlook carry context between applications, while new connectors and a Moody’s MCP app provide access to market, research, credit, and company data. The announcement says the updates pair best with Claude Opus 4.7, which leads Vals AI’s Finance Agent benchmark at 64.37%, and states that templates are available today through the financial services marketplace.
Inkling model from Thinking Machines Lab now on Databricks
Databricks announces that Inkling, Thinking Machines Lab’s first open-weights model, is available to enterprise customers through the Unity AI Gateway. The model is positioned for coding and agentic reasoning workflows, supports multi-modal inputs, and can be applied to enterprise data, including proprietary codebases, internal documentation, and domain-specific data. Unity AI Gateway provides centralized security, permissions, audit logging, policy enforcement, cost controls, budgets, and observability, while data remains within the governed environment; Inkling is invoked through a REST API, with SQL query support planned. Teams can try it in AI Playground, deploy a governed endpoint, connect coding agents such as Cursor, OpenCode, or Pi, and build agents with Agent Bricks, with open weights enabling customization and inference-cost optimization without per-token API pricing.
Replacing the HDB: ClickHouse for historical ticker data
Historical market-data systems separate a durable archive from the real-time trading path, but the archive still must support research, backtesting, transaction-cost analysis, and regulatory reconstruction. The worked example uses Binance’s public archive to load trades and quotes for five USD-M perpetual-futures symbols into ClickHouse, with a MergeTree table ordered by (symbol, ts) and codecs including DoubleDelta and ZSTD. SQL queries calculate VWAP, hourly OHLC candles, and as-of joins, while ClickHouse’s sparse primary index skips most granules for symbol-and-time filters. Across ten months, 16.2 billion rows compress to 76 GiB on disk; the post reports a ClickHouse Cloud cost of $1.88 per month and argues that typical symbol-and-time queries retain low latency at that scale, while broader scans read more data.
Building Faster Cryptography with Carryless Multiplication in NVIDIA CUDA 13.3
NVIDIA CUDA 13.3 introduces clmad, a native PTX carryless multiply-accumulate instruction for Ampere and newer GPU architectures. Prior to this release, GPUs lacked dedicated hardware for carryless multiplication, requiring developers to emulate binary extension field arithmetic with ALU operations or bitsliced circuits. The new primitive performs 64-bit by 64-bit carryless multiplication into 128-bit results with accumulation, allowing full 128-bit field multiplications via Karatsuba in six instructions. Benchmarks demonstrate that clmad achieves up to 6.3 TB/s throughput for GHASH on the NVIDIA B200, representing an 18.8x improvement over bitslicing. Additionally, sum-check protocol latency in zero-knowledge proving workloads decreases by up to 13x on the B200 and 4x on the RTX 5090.
AI-Enabled Advisory Services for Higher Education
Higher-education call centers face costly, limited-coverage monitoring of advisor conversations and brittle, slow methods for identifying student concerns from transcripts. The proposed workflow deploys OpenAI Whisper on Databricks Model Serving, applies AI Functions for sentiment, topics, intent, and rubric scoring, and uses Unity Catalog to govern the resulting data. For advisor quality, an LLM-as-a-judge evaluates every transcript against a reference-table rubric, returns a weighted 1–5 overall score and per-criterion scores, and routes flagged calls for targeted QA review instead of random sampling. For student insights, quarterly transcript enrichment feeds an Agent Bricks Knowledge Assistant for cited reasoning over raw calls and a Genie Space for structured trend queries, while LangGraph orchestrates UC SQL functions as tools. Together, these components let non-technical advisors, mentors, and QA managers query student interactions without reaching out to a data SME.
Data-Native AI Agents: Why Agents Must Move to Your Data
Enterprise AI pilots often move data into separate vector databases, SaaS LLMs, or serving layers, creating governance gaps, compounded latency, fragmented costs and observability, and duplicated lifecycle work. The post advocates data-native agents: models, agents, tools, retrieval, and memory run inside the governed data platform, with policy enforced during query planning and computation rather than after responses are produced. It argues that post-hoc controls cannot undo sensitive information encoded in aggregations and can trigger token-burning retry loops. For state and memory, it presents Lakebase, managed PostgreSQL within Databricks, as transactional storage and a shared source of truth for multi-agent swarms. The described platform pattern combines Unity Catalog, Unity AI Gateway, Model Serving, MLflow 3, AI Search, Lakebase, and business-context services, and recommends inventorying workloads already outside the perimeter before closing seams incrementally.
Model Routing Is Simple. Until It Isn’t.
Model routing in agentic systems is presented as a systems optimization problem rather than a simple classification task. On 417 AppWorld Test Challenge tasks with the same CodeAct agent, Claude Sonnet 4.6 cost $79 total ($0.19 per task), versus $155 ($0.37 per task) for GPT-4.1, despite GPT-4.1 having lower listed token prices; cache-read pricing and repeated context changed the effective result. The discussion also shows why task difficulty and model speed are incomplete signals: hidden execution work, specialization, governance constraints, routing overhead, cache state, hardware, and endpoint load all affect decisions. The described router optimizes cost, quality, and latency together, producing configurable cost-accuracy operating points; one latency-oriented configuration reached 84% accuracy for $93 and 83 seconds, while adding about 6 ms and 2 kB per task.
Exploring Hierarchical Interest Representation For Meta Ads Deep Funnel Optimization
Hierarchical Interest Representation is an upstream representation layer for Meta Ads that addresses sparse deep-funnel feedback by connecting users’ inferred interests with advertisers, products, services, and other ads entities. It models a typed, heterogeneous engagement graph with a transformer architecture, sparse attention for long-range relationships, bias-aware attention, self-supervised cross-view distillation, and multimodal content features processed through vision or language models. The system projects the raw graph into learned interest primitives and cascading super-graphs, producing universal embeddings and Bag-of-Meaning tokens at multiple granularities. Temporal cutoffs and chronological splits prevent future-information leakage, while a pipelined online graph engine reportedly achieved a 30x wall-clock speedup over a synchronous baseline during testing. Trained end-to-end on billions of interactions, the representations are designed for personalization, retrieval, supervision, and ranking integrations including GEM, Andromeda, and the Adaptive Ranking Model.
Anthropic commits $10 million to Canadian AI research
Anthropic is committing $10 million CAD to Canadian research institutions to support beneficial and responsible AI applications, building on Canada’s contributions to neural networks, deep learning, and reinforcement learning. The announcement names partnerships with Amii, Mila, Vector, CHEO, CAMH, Université Laval, University of Toronto, and the University of Saskatchewan, with Claude credits supporting work spanning trust and safety, health, scientific discovery, mental health, robotics, sustainability, and public services. Amii, Mila, and Vector will also join Anthropic’s for Startups program, providing affiliated Canadian startups with at least $5,000 USD each in API credits. A Canadian country brief based on the March 2026 Anthropic Economic Index reports that Canada ranks eighth in Claude.ai use, accounts for 2.6% of global consumer use, and ranks second by per-capita adoption; usage is highest in British Columbia and closely follows professional, scientific, and technical employment.
Expanding the Uber Transit Innovation Fund: meet our next cohort of recipients
The Uber Transit Innovation Fund responds to reduced transit funding by supporting public agencies and partners pursuing projects that keep communities moving. Launched with a $1 million commitment, the fund initially offered up to $50,000 to each of as many as 20 forward-thinking US transit agencies, municipalities, and regional authorities. Strong interest from US and Canadian agencies led to an expanded investment and eligibility, following an initial cohort of nine winners. The additional 26 recipients bring total awarded funding to $1.7 million, or $1,705,000, with projects spanning microtransit in Oregon, food access in Richmond, Virginia, and transportation gaps across urban, suburban, and semi-rural communities. Additional funding and recipients are scheduled to be announced by the end of the month.
Inkling by Thinking Machines now available on Modal
Thinking Machines released Inkling, a general-purpose multimodal model that accepts text, image and audio inputs and generates text outputs. The 975B-parameter mixture-of-experts transformer activates 41B parameters per token, uses a 1M-token context window, and employs local attention: five sliding window layers for every global layer. Modal provided day-zero support with a custom DFlash block-diffusion speculator, achieving 250 tokens per second per user on 8x B200s with 67% faster interactivity than the model’s built-in multi-token prediction path. The DFlash drafter uses all local attention and causal layers, keeping cost flat as block size grows. Inkling is available today as an OpenAI-compatible shared endpoint through Modal Auto Endpoints with token-based pricing, allowing users to try the model interactively.
Together AI brings Thinking Machines Lab’s new model Inkling on day 0
Thinking Machines Lab released Inkling, and Together AI announced day-zero access through its production inference platform. The multimodal mixture-of-experts model accepts text, image, and audio inputs and generates text through a unified decoder architecture, with controllable inference effort for trading reasoning depth, token usage, and latency. Inkling has 975B total parameters, 40B active parameters per token, a 1M-token context window, query-conditioned relative attention, short causal convolutions, and shared-sink MoE routing; Together serves it with a FlashAttention-4-based kernel. Preliminary evaluations report strong results across scientific reasoning, mathematics, coding, agentic, vision, and audio benchmarks, while the model is also post-trained for forecasting and calibrated prediction. It is available on Together Serverless through OpenAI-compatible APIs without infrastructure provisioning.
New in Together GPU Clusters: Reliability and control for production GPU clusters
Together GPU Clusters introduces changes for production training and inference workloads, addressing hardware failures, scheduler problems, and the limits of single-admin access. Platform health improvements include continuously running passive health checks, human-approved node repair recommendations, and Slurm-on-Kubernetes 2.0, which adds self-healing daemons, durable job accounting, descendant-process cleanup, and refreshed GPU state after reschedules. Operational controls include a cluster details view with health, utilization, events, configuration, and node actions, plus external OIDC, startup scripts, and an acceptance-test opt-out. Newly provisioned Slurm clusters use the latest stack by default, while existing managed clusters can be migrated during a maintenance window at no cost; the stated result is faster recovery, fewer support tickets, and greater visibility and control for growing teams running large clusters.
Welcome Inkling by Thinking Machines
Thinking Machines Lab has released Inkling and the smaller Inkling-Small family as open multimodal models that accept image, text, and audio inputs. Inkling is described as a roughly 1T-parameter decoder-only Mixture-of-Experts model, with 975B total and 41B active parameters, a 1M-token context window, and training on 45 trillion multimodal tokens. Its architecture combines relative attention, alternating global and sliding-window attention, short convolution, shared experts, hierarchical image patchification, and discretized mel-spectrogram audio processing. The release includes BF16 and NVFP4 checkpoints, speculative MTP layers, day-zero support in Transformers, SGLang, vLLM, and llama.cpp, plus deployment configurations and interactive voice and image demos. Reported benchmarks show Inkling and Inkling-Small achieving strong but varied results across reasoning, coding, vision, audio, factuality, chat, and safety evaluations.
Top topics
- Architecture834
- Performance816
- LLMs660
- Machine Learning622
- Open Source563
- AI Agents553
- AI470
- Developer Experience448
Weekly digest
The best engineering reads, once a week.