What shipped in AI engineering: October 2026
380 posts from 13 sources, grouped by area and ranked by relevance to engineers shipping production AI.
Models & LLMs
Model releases, evaluations and LLM research · 101
- The Hidden Cost of LLM Routing: Why Switching Models Costs More
The article investigates LLM routing and finds that switching models during a session can negate the cost savings expected from using cheaper inference models.
- FairStress: A deeper evaluation of fairness in LLM decision-making
FairStress is a new benchmark that evaluates demographic fairness in LLM decision‑making by probing how persuasive “adversarial prefixes” affect a model’s willingness to change a decision across demographic groups.
- Nvidia expands Nemotron LLM family with hyper-efficient models for agentic systems
Nvidia announced three new Nemotron 3.5 models—including the customizable Nemotron 3.5 Super LV for always‑on AI agents and the edge‑focused Nemotron 3.5 Bolt for local hardware—designed to span deployments from large enterprise servers…
- NVIDIA Open Models Advance Europe’s Sovereign Intelligence Across Industries
NVIDIA announced an expanded portfolio of open‑weight models—including Nemotron 3.5 Super VL, Nemotron 3.5 Bolt, Nemotron Retriever, Kumo Structured, Kumo Engine, Cosmos 3.5, Alpamayo 2 Edge, BioNeMo, and Ising—to let developers and…
- Mistral releases open weights for Mistral Large 4, providing sovereign control for the company’s ‘Le Chonk’ AI model
Mistral AI has released the open weights for its newest frontier model, Mistral Large 4 (“Le Chonk”), a 1 trillion‑parameter LLM with 52 billion active parameters and native multimodal capabilities.
- Jev Your Engines: Seven Open Decision Models vs. a Generative LLM
The article presents a hands‑on comparison of seven open decision‑making models against a generative large language model, evaluating routing accuracy, latency, and output reliability.
Show 95 more in Models & LLMs
- Closing the Binary Blind Spot: Domain-Knowledge Injection for LLM Malware Triage in npm Packages
- Inception says Mercury 3 beats GPT-6 Luna and Claude Haiku 5.5 at up to 8x the speed
- Nvidia releases all-in-one adaptable Nemotron 3.5 audio transcription and translation model
- Nvidia, Dell back $200M round for AI model developer doubleAI
- The hidden threat in Anthropic’s prospectus: diseconomies of scale
- Anthropic files S-1 paperwork for blockbuster AI IPO
- Trump administration orders ban on DeepSeek in federal agencies
- Knowledge graphs fuel AI innovation for financial services at Raymond James Financial
- Fine-tune a search agent with multi-turn RL on Amazon SageMaker AI
- Mistral launches open-source Mistral Large 4, details AI roadmap
- Reflection AI open-sources Beam’s weights, post-training recipes
- AI startup doubleAI debuts Maxwell ‘code execution model’
- Amazon SageMaker JumpStart: 2026 year-to-date in review
- EmbeddingGemma 2: an open, lightweight multimodal embedding model
- Deploy decision models with llama.cpp on Amazon SageMaker AI
- Scaling Indic Pretraining: 2.8 Trillion Tokens and an Open Data Generation Pipeline
- Reflection AI debuts open-source Beam model with 501B parameters
- Otter.ai raises $320M in funding for its AI meeting assistant
- Building Reasoning Models from the Ground Up: A Full Video Course
- Run multi-node inference workloads using Ray LLM Deep Learning Containers
- Gemini 4 Carbon: Our next frontier model
- Build specialized multimodal models with LFM3
- Welcome Muse Spark Open: open-weight frontier multimodal mixture-of-experts model from Meta
- Google DeepMind and Singapore Government collaborate to build fine-tuned AI model for scam detection
- Anthropic releases Claude Haiku 5.5 small model and halves Sonnet 5.5 cache read prices
- Fine-Tuning vs. Frozen Embeddings: When to Adapt the LLM vs. the Sklearn Head
- How NVIDIA GPUs Help Accelerate OpenAI’s GPT-6 Astra Ultrafast
- From spot fixes to measured gains: improving document extraction accuracy with ai_extract and MLflow Prompt Registry
- Strengthening AI safety evaluation through independent third-party review
- DeepSeek releases updated V4.1-Flash model it says completes tasks faster
- Ai2 releases Olmo-core 3 to make developing large mixture-of-experts LLMs more efficient
- Same Answer, Different Reason: What a Reasoning Model Says When It Follows a Planted Hint
- Beyond the KV cache: Training and serving i14’s Laplace 3B linear attention model on AWS Trainium2 with native PyTorch
- Measuring the Creativity Potential of LLM Agents
- Google debuts flagship Gemini 4 Carbon model with cybersecurity upgrades
- Welcome to Google Gemini 4 Argon
- OpenAI to make its most advanced AI models available in Google Cloud
- Granite Switch 4.1.1 — Update models like you update software
- Is the #1 AI Model Really #1?
- Welcome Le Chonk: Mistral Large 4 on Hugging Face
- Introducing AnySpeechRecognizer and AnyDecisionModel: Open Models for Speech and Decisions on Apple Platforms
- The sleep framework: A new way for LLMs to retain knowledge in continual learning
- Can we trust LLMs to judge whether a research idea is novel?
- Bringing foundational models to the front lines of TB screening
- How Specializing the Vocabulary of SLMs Can Cut Inference Cost by 77% in Real Applications
- TorchNeuron: running HF models on AWS Trainium is easier than ever
- Welcome Beam: a 501B open model you can own, customize, and scale
- Developing Gemini 4 Argon responsibly and safely
- Gemini 4 Argon: Built to solve your hardest problems
- Unlocking Earth AI’s planetary geospatial foundation models for global public health
- Cutting both ways: AI sharpens systems protection while opening new exploits for malicious actors
- Can we change an AI model’s architecture without retraining it from scratch?
- Deepgram targets voice AI on air-gapped laptops
- Microsoft reveals cutting-edge MAI-Base-2 model with 2T parameters
- Speculative Decoding Explained: How to Get 3× Faster Inference from Any LLM
- Building a Custom LLM Evaluation Harness in Python from Scratch
- Mixture-of-Experts kernel engineering on Amazon SageMaker AI
- Your Six AI Models Are Really Two: The Math of Correlated Mistakes
- AI Engineers, Applied Scientists, Forward Deployed Engineers: Let's Unpack.
- The Consistency Quadrant: A Visual Guide to LLM Reliability
- Claude Code Mods: Custom Hooks to Fix My Biggest AI Coding Frustrations
- Speeding up language models for travel applications
- Q&A: How AI came to model the “messy” human mind
- Ai2 releases OlmoEarth v1.3 state-of-the-art AI-ready satellite embeddings for the whole planet
- AI Harms and Their Technical Causes
- The Number Your Quantization Gate Doesn't Have
- AI model decodes the language cells use to communicate
- Why Telecom Operators Are Building Their AI Strategy on Open Models
- Genie 4: A new generation of world simulation
- Moonshot AI launches new flagship Kim K3.1 Pro model with 3T parameters
- Why Parameter Count Is Not Model Complexity: An Introduction to Effective Dimension
- The Fraud Analyst AI Coworker: Detecting Telecom Fraud in Real-Time on One Platform
- Building and Visualising Kolmogorov–Arnold Networks
- Data Enablement for AI: A Practitioner's Guide to 6 Advanced Architectural Patterns
- MPS in PyTorch 2.15
- Research@ 2026: 10 ways we're amplifying human ingenuity
- Does better work always mean better workers?
- Google introduces cutting-edge Genie 4 world model
- Supercharge regulated workloads with Claude Code and Amazon Bedrock
- How to Build Bug-Free Applications with Claude Code
- Why Temperature 0 Isn't Deterministic
- Binning or Feature Selection: Which One Won Us an Insurance Hackathon?
- Making machines that can listen
- Synthetic-to-Real Transfer with Frozen DINOv3 Features
- Can Conformal Prediction Take You from Accurate to Reliable Machine Learning?
- Giving AI a “crystal clear” view of the materials world
- How to avoid AI’s ‘valley of despair’
- Could On-Policy Self-Distillation Revolutionize Post-Training?
- Wabi reels in $55M for its consumer-focused AI assistant
- Better speech synthesis for audiobooks
- Researchers improve weather forecasts looking two to six weeks out
- How artificial neural networks build their own learning machinery
- 3 Questions: What is the best path forward for AI in academia?
- Skewness and Kurtosis: From Moments to Machine Learning
- Google X co-founder Sebastian Thrun raises $130M for AI engineering startup Dulo
Agents & MCP
Agent frameworks, tool use, MCP and orchestration · 120
- Introducing Reflection Beam: an open-weight model for agentic coding on AWS
Reflection AI unveiled Beam, a 501 B‑parameter sparse mixture‑of‑experts model (23 B active parameters) designed for agentic coding and long‑horizon software tasks, now released with open weights on Hugging Face under Apache 2.0.
- Adaptive agentic intelligence framework using Strands Agents
The Adaptive Agentic Intelligence Framework (AAIF) uses a 1‑billion‑parameter llama3.2 SLM as a lightweight orchestrator to classify and route requests between local Ollama‑hosted models (including llama3.2‑vision) and cloud‑hosted…
- NVIDIA Cloud Partners Scale Across Europe — Sovereign, Open and Built for Safe AI Agents
NVIDIA’s European Cloud Partners are expanding sovereign, open, and agent‑centric AI infrastructure, leveraging the DSX AI Factory platform to boost compute efficiency.
- The race to deploy AI agents finds a familiar hurdle in data access
Enterprises deploying AI agents are finding that role‑based access control (RBAC), rather than the choice of model, determines whether agents can be trusted in production.
- NVIDIA Agent Toolkit Expands, Grounding Enterprise Agents in Company Data
NVIDIA expands its open‑source Agent Toolkit with the Agent Knowledge Foundation, adding CUDA‑X libraries, Auto‑Ontology, Nemotron Retriever and the Kumo family of structured models to let enterprises ground agents in proprietary data…
- MiniMax launches M3.1 Flash model, says it cuts AI agent costs by 70%
MiniMax Group Inc. released the M3.1 Flash large language model, a faster and cheaper successor to its June‑launched M3 model, claiming it can reduce AI‑agent operating costs by 70%.
Show 114 more in Agents & MCP
- How MoEngage turned over 400 APIs into agent tools with AgentCore Gateway
- Securing AI agent identity with X.509 PKI on AWS
- From data to decisions: Conversational supply chain intelligence with Amazon Quick and Amazon Bedrock AgentCore
- Accelerating software delivery with agentic QA automation using Amazon Nova Act – Part 3
- Build agentic retrieval on a Managed Amazon Bedrock Knowledge Base with CDK
- Work management context for AI agents with Smartsheet MCP on AWS
- Europe’s Industrial Leaders Reinvent Engineering and Manufacturing With NVIDIA
- Building intent-driven UI testing with Amazon Nova Act
- Agentic AI Foundation adds four more open-source AI tools
- Build a semantic layer for agentic AI on AWS with AtScale and Amazon Bedrock AgentCore
- Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models
- GTC Berlin 2026: Live Updates From NVIDIA CEO Jensen Huang’s Keynote
- Cerebras positions inference speed as a competitive edge in agentic AI
- European Cities Deploy NVIDIA-Powered AI Agents to Run Smarter Urban Operations
- Build a semantic layer for agentic AI on AWS with Strategy Mosaic and Amazon Bedrock AgentCore
- Exclusive: Iterate.ai’s Lifeboat runs up to six times more AI agent sessions per GPU
- AI agents cut critical research time for humanitarian teams in conflict zones
- Virtana lets its AI agents pull in agents from other vendors to work IT incidents
- Build a semantic layer for agentic AI on AWS with Denodo and Amazon Bedrock AgentCore
- Crux Labs raises over $6M to improve enterprise agents with human judgment
- Agentic retrieval with LangChain and Amazon Bedrock Knowledge Bases
- NVIDIA Advances Agentic Security for Media, From European Public Service Broadcasting to Global Content Production
- The agentic AI era is forcing security and resilience into a single discipline
- Evaluating multi-agent systems for explainability and helpfulness with Amazon Bedrock AgentCore
- Groundhog Day for AI agents: Prompt caching patterns for using OpenAI models on Amazon Bedrock
- Dell’s security chief warns AI is lowering the barriers to sophisticated cyberattacks
- Discover tools at runtime with AWS Agent Registry
- Accelerate agentic AI with NVIDIA Nemotron models on Amazon EKS and SageMaker HyperPod EKS
- Building trust in AI agents: A production-ready evaluation framework with Amazon Bedrock AgentCore
- Meta’s ads MCP server comes to Databricks: Put your customer intelligence to work in advertising campaigns
- Anaconda expands beyond Python with agent swarms and AI security testing
- SAP expands Joule into an agentic work layer as Autonomous Enterprise goes live
- Transforming the data path: Dell’s new releases move enterprises closer to the agentic data center
- Regal debuts Goal Oriented Agents, putting an actor and a director model on every call
- Build agent memory with NVIDIA NeMo Agent Toolkit and Amazon S3 Vectors
- Workflow context for AI agents with Smartsheet MCP on AWS
- Google Cloud introduces Gemini agent to change enterprise work
- SAP advances its autonomous enterprise push with smarter, governed Joule agents
- Add secure Web Search to Claude Desktop with Amazon Bedrock AgentCore
- I Told an AI Agent That Its Wrong Answers Were Right. One Model Believed Me.
- Motive brings agentic AI to physical operations with agents that act before problems escalate
- Pax8 launches 8os to let IT partners govern AI agents and 8bot as a managed AI co-worker
- Building an Agentic AI Platform on Amazon Bedrock AgentCore
- How to Add Guardrails to a Local LLM Agent
- Atlassian lays groundwork for humans and AI agents to work side by side
- Your Agent’s Tools Can Make Decisions, Too
- Trust in Depth: Giving agents more autonomy without losing control
- Midwest Wheel builds toward AI agents that fix problems
- New agent skill: Amazon SageMaker optimized generative AI inference for your coding agent
- IBM allows on-prem deployment of its Bob agentic development platform
- AWS debuts Strands Decider 2B, a first lightweight decision model for accelerate agentic workflows
- 5 production challenges in multi-agent systems and how Deep Insight solved them
- AI governance shifts to runtime as always-on agents spread
- Phasis develops platform for agentic orientation using World APIs and reality layers
- Lakebase and Agentic SDLC: Branching Databases for Coding Agents
- Scaling cloud migrations with agentic AI on Amazon Bedrock AgentCore
- An AI Agent Can Spot a Mistake and Still Not Fix It
- Agentic AI in Production: Why Your Agent Keeps Answering After Its Tools Fail
- Anthropic introduces Claude Duet, a proactive AI agent for professionals
- Agentic workloads break assumptions about software testing. Here’s how to cope
- Nvidia, Samsung back $90M round for AI agent startup Nous Research
- Turning Database Grading into RL Rewards: Post-Training Agents with ThinkingBox
- Grounding AI agents in biological reality: Building blocks for autonomous wet labs
- Putting Earth AI to work: Geospatial agents for our planet's biggest challenges
- Run fast, private agents on-device with LFM2.5-24B-A2B
- Gemini 4 Argon for Robotics
- The model that didn't exist, so you made it yourself
- The Agent Said It Was Done. The Database Disagreed.
- Built for Agents: The Case for Spark Declarative Pipelines
- AI work assistant startup Town secures unicorn status after raising $108M just weeks after launching
- AI Agent Benchmarks Explained: SWE-Bench, GAIA, WebArena, and More
- Can TypeSafe's Jev Make AI Agents Safer Without Another LLM?
- Build self-improving document processing using Amazon Bedrock AgentCore
- We Specified an AI Agent in 2016. We Called It Analytics.
- Automation Anywhere acquires Boost.ai to expand customer-facing voice AI
- Sweep thousands of leases for compliance using Amazon Quick and the Adjudicated Query pattern
- nemotron-halloween
- Preview web apps live as your coding agent builds them with Amazon Bedrock AgentCore runtime
- 7 Ways Your AI Agent Is Wasting Tokens (And What to Do Instead)
- Beyond hours saved: Building the business case for agentic automation
- Build a repeatable PHP-to-Java migration engine with Kiro
- Startup nFox to open-source its ‘sandbox’ for open AI models
- OpenAI acquires Orchestra Bio to bring AI agents into drug development
- Introducing Agent Memory on Databricks
- Things That Go Beep in the Night: Agentic AI Joins the Search for Extraterrestrial Signals
- Jedify formally launches Data Agents for building AI analysts in plain language
- Bringing the Spark to Europe: Local Intelligence Powers Next Wave of Agents
- Dell sees company knowledge, not AI models, as its competitive edge
- NVIDIA, Microsoft Kick Off a New Beginning for Windows PCs With RTX Spark and AI Agents
- Building a context-aware AI assistant on AgentCore and OpenClaw
- Graph-centric agentic intelligence
- Red Hat and Nvidia expand their partnership to secure autonomous AI agents
- MemCurrent: Designing Persistent Memory for Production Agents
- Synchronous vs. Asynchronous Agent Execution: Architecture Patterns for Production
- Multi-Turn Reinforcement Learning on Amazon SageMaker AI: A hands-on guide
- Who Watches the AI?
- Startup Action State launches tools for verifying interactions between AI agents
- How Tamnoon built AI-powered cloud security remediation with Amazon Bedrock
- SolarWinds’ Avior pairs AI agents with deterministic execution for enterprise IT
- Speeding up planning decisions with AI
- Accelerating the search for antimicrobial peptides with Google Antigravity
- Analyzing technical documents with Deep Review
- CoreWeave Forge aims to speed up the AI improvement loop
- How (and Why) to Build an AI Agent from Scratch in Python
- How to Build a Control Plane for AI Agents
- Making Amazon Quick enterprise-ready: Automated, auditable cross-account resource promotion
- Monday.com sees human-AI collaboration as the future of work
- Choosing the Right Agentic AI Framework for 2026: A Decision-Tree Approach
- How Worldwide Express automated freight exceptions with Amazon Bedrock
- How the UK Amazon Quick User Group grew in 2026
- Uvi raises $7.5M to help AI agents learn business context and run at eight times lower cost
- Building AI builders: Playbook for closing the AI knowledge-capability gap
- Voting on Features: Federated Feature Selection Without Sharing Data
- The Pulse: RoR creator sparks new “death of coding by hand” debate
Inference & infrastructure
Serving, latency, GPUs and deployment · 101
- Push-based KV transfer to improve TTFT and TPOT for disaggregated inference in vLLM
vLLM’s new push‑based NIXL connector moves KV‑cache transfer off the critical path of disaggregated prefill/decode inference, cutting time‑to‑first‑token (TTFT) by up to 3× and improving time‑per‑output‑token (TPOT) by up to 30 % for…
- Stream low-latency speech from AI agents using Kokoro TTS on Amazon SageMaker AI
This post demonstrates how to achieve sub‑second latency for voice agents by streaming each generated sentence to a hosted Kokoro‑82M text‑to‑speech model on Amazon SageMaker AI, allowing audio playback to begin before the full response…
- Pioneers Across Europe Deploy NVIDIA Earth-2 Open Models to Power Real-World Weather and Climate Prediction
NVIDIA Earth‑2 now includes StormScope EU, a regional short‑term forecasting model trained on European weather data that ingests EUMETSAT satellite observations across 16 Meteosat channels and outputs cloud predictions every 10 minutes.
- NVIDIA Confidential Computing Expands, Enabling Frontier Model Builders to Securely Scale Across Global Infrastructure
NVIDIA announced the Confidential Bare Metal deployment option for its Confidential Computing stack, allowing frontier model builders to run sensitive workloads on dedicated third‑party bare‑metal infrastructure while retaining full…
- Deploy RFD4-Proteina on Databricks
The announcement describes how to bring the RFD4-Proteina model to research data using AI‑assisted development on Databricks with serverless NVIDIA GPU compute.
- NVIDIA Nemotron 3.5 Transcribe and Granary v2: Intelligent Multilingual Transcription You Can Prompt and Adapt
NVIDIA Nemotron 3.5 Transcribe is an open‑weight, promptable multilingual speech model that jointly performs transcription, speaker diarization, translation, and audio Q&A across 40 languages (31 for translation) and supports up to…
Show 95 more in Inference & infrastructure
- Improve TTFT for multi-turn agentic workloads with bi-directional KV transfer in vLLM
- Exowatt beats frontier models on accuracy with SageMaker serverless model customization
- Thin, Light and Built for Everything: NVIDIA RTX Spark Windows Laptops Gain New Models, Agent Tools and Games
- How Deepgram optimized voice AI at scale with AWS Trainium
- Stop guessing auto scaling targets for TTS on Amazon SageMaker AI
- NVIDIA Open Models Power a New Wave of European Public Sector Projects
- Distributed training with Ray Train DLCs and EFA on EKS
- NVIDIA Transforms European Retail, From Digital Discovery to Physical Operations
- Leading European Financial Services Organizations Embrace Open Models to Build Proprietary Intelligence
- Lend and borrow resources across teams with AWS Batch’s quota management for SageMaker Training jobs
- Announcing SOCI-indexed images for faster SageMaker AI jobs
- How Companies Are Using NVIDIA DSX to Break the Power Bottleneck
- DeepSeek open-sources DeepInfer, the inference framework behind its production services
- Fine-tuning GPT-OSS-20B for multilingual reasoning on Amazon SageMaker HyperPod
- How FactSet cut model training and deployment from days to hours with Databricks AI Runtime
- European Leaders Turn to NVIDIA Vera CPUs to Build Local Infrastructure
- Hybrid infrastructure decisions put AI costs and control in focus
- NVIDIA Expands Advanced Technology Access for Europe’s Scientific Community Through Open Learning Resources, Collaborations and Compute
- Cloudera expands its AI ecosystem with partnerships and packaged blueprints
- How Hamlet uses Databricks Lakebase Postgres to test changes against production data
- Oracle brings Google’s TPUs to its cloud, adds ChatGPT integrations
- The Nordics and NVIDIA Show What’s Possible When a Region Commits to Advanced Computing
- CoreWeave targets GPU utilization in continuous AI post-training
- Google unveils TorchTPU in a push to break Nvidia’s stranglehold on AI compute
- IBM-backed llm-d framework adds controls for running AI models in-house
- Antseed launches a decentralized marketplace for AI inference that dramatically lowers developers’ costs
- NetApp’s Novus tackles metadata bottlenecks in AI data infrastructure
- Vast Data challenges fragmented AI pipelines as data sovereignty demands grow
- Spain-Based Startups Advance Regional Innovation With Barcelona Supercomputing Center
- NVIDIA Announces New Investments in the UK Startup Ecosystem
- Germany Wants to Lead the World in Industrial AI: How NVIDIA Will Help
- How uniopen customized Amazon Nova to their retail moderation policies for production deployment
- Reducing Qwen training time by 6x on AI Runtime
- Developers can now access SageMaker Studio through the new AWS experience
- NVIDIA Announces New Investments in France’s Startups
- Deploy and evaluate models with Amazon SageMaker JumpStart
- All Systems Go: Global PC App Ecosystem Ready for NVIDIA RTX Spark
- Dell and Intel weigh the costs of enterprise AI scalability
- Willow taps CoreWeave to simplify AI model training
- Liquid AI builds personal AI around device-level context
- Nvidia ties AI factory economics to tokens and power efficiency
- Announcing new ways to customize models with SageMaker AI, without managing GPUs
- NVIDIA Cosmos Opens New Tools to Train, Evaluate and Deploy World Models
- RTX Spark Ignites Global Celebration as New Windows PCs Arrive
- CoreWeave targets AI inference bottlenecks with full-stack optimization
- How to teach with Databricks Free Edition
- Does Your Model Run on AWS Trainium? Look It Up
- Ensemble forecasting with IBM Granite time-series models
- AnyLanguageModel 1.0: One Year of Local Models on Apple Platforms
- AI Agents Beat PyTorch: Writing Faster CUDA Kernels
- Write your own training loop — without managing GPUs — with Amazon SageMaker training session APIs
- OCP Global Summit: Amid Ecosystem Adoption, NVIDIA Opens Vera Rubin Architecture to Help Scale AI Factories Worldwide
- Always-on AI agents turn infrastructure into a continuous learning loop
- pgvector vs. Dedicated Vector Databases: A Production Benchmark
- Marvell’s early chipmaking gamble pays off amid AI demand
- Build a voice travel concierge with Amazon Bedrock AgentCore, Managed Knowledge Base and Nova Sonic
- NetApp and Nvidia rethink storage for AI factories
- Train Qwen3 on AWS Trainium with torchtitan
- Databricks AI Runtime and NVIDIA make it easier to customize open models with enterprise data
- Nvidia reportedly in early talks to buy, acquihire or invest more in Reflection AI
- Share GPU clusters across teams with isolation and fairness using Amazon SageMaker HyperPod
- Geospatial queries are up to 38X faster with Lakehouse//RT
- Scaled Cognition targets customer service with low-cost voice model
- How our AI Research team uses AI Runtime
- The science behind TorchNeuron, Amazon’s new PyTorch backend for AWS Trainium chips
- CoreWeave makes the case for an open, full-stack AI cloud
- NVIDIA DGX Spark 64GB Gives Developers More Ways to Build and Scale Local AI
- Uplifting conversion across the acquisition funnel with personalization using contextual bandits on AWS
- Empowering frontline maternal care with AI-assisted ultrasound
- Profiling TorchNeuron (Part 1): System-level profiling on AWS Trainium.
- Hugging Face ❤️ TorchTPU: TPUs, Natively
- Welcome FLUX.3 Image Klein in Diffusers
- Introducing Falcon ASR
- Toward provably private learning from federated data
- Seismora builds a control plane to route AI workloads across devices and clouds
- Kazakhstan Expands Compute for Its AI Future
- NVIDIA cuDSS Now Generally Available to Accelerate Engineering and Scientific Simulations
- AI’s production push puts data platforms on the hook for results
- AI Runtime is Generally Available
- Google debuts SynthID Detector tool for flagging AI-generated content, but it’s far from perfect
- Smarter cost controls for serverless
- The Pulse: Anthropic builds its own source control system
- Fleet-wide computer vision lifecycle at the edge with SageMaker SDK v3
- Orchestrating SageMaker AI endpoint auto scaling with AWS CDK
- NVIDIA Unveils Minerva to Bring End-to-End Observability for AI Factories
- Google Cloud and NVIDIA Announce General Availability of NVIDIA Vera Rubin Infrastructure
- Securing Quick access from corporate networks with AWS PrivateLink
- How AI Is Changing the Pace of Discovery for Millions Living With Dementia
- Using AI to mitigate the growing environmental threat of data centers
- Supercomputing researchers document evolution of AI hardware
- Best practices for Amazon SageMaker HyperPod administration and governance
- Manage Amazon SageMaker HyperPod Spaces directly from SageMaker Studio
- From Scan to Treatment Plan, AI Helps Close Breast Cancer’s Deadliest Gaps
- Productive, Durable, Fungible: How NVIDIA AI Factories Maximize Return on Investment
- AI infrastructure management software startup ASI hits $3B valuation after raising $325M in funding
RAG & retrieval
Retrieval, vector search and knowledge bases · 12
- Your RAG Pipeline Is Missing a 1994 Algorithm
The article highlights that typical RAG pipelines rely on embeddings that capture semantic similarity (e.g., “broken” ↔ “defective”) but may miss exact string matching nuances such as distinguishing “ZX-4821” from “ZX-4812”.
- Building a RAG Pipeline with LLM Embeddings and Scikit-learn
This article demonstrates how to construct a lightweight, end‑to‑end Retrieval‑Augmented Generation (RAG) pipeline by leveraging scikit‑learn for vector similarity search and open‑source language models to generate embeddings.
- Choosing a Chunking Strategy for Your LLM Apps: A Decision-Tree Approach
The article presents a decision‑tree framework to help engineers select the most appropriate chunking strategy for retrieval‑augmented generation (RAG) systems.
- NEAREST BY Join: Scaling Vector Search in Databricks Runtime
The article introduces **NEAREST BY Join**, a technique for scaling vector search within the Databricks Runtime, aiming to improve the efficiency of similarity lookups for applications such as chatbots.
- Building live intelligence: Databricks AI Search supports Streaming Tables, Materialized Views, and Managed tables using Iceberg
Databricks AI Search now natively supports Streaming Tables, Materialized Views, and Managed tables built on Iceberg, extending its retrieval capabilities to real‑time and curated data sources.
- Does Agentic RAG Actually Retrieve Better Evidence? I Tested It on 298 Questions
The article evaluates an agentic Retrieval‑Augmented Generation (RAG) system by running it on 298 questions, allowing the AI agent to re‑search whenever it chose.
Show 6 more in RAG & retrieval
- Evaluating Graph-RAG vs. Standard RAG: A Hallucination Benchmark on Fact-Dense Queries
- When Labels Say No, But Embeddings Say Yes
- Adding Temporal Reasoning to Graph-RAG: Tracking Fact Freshness and Staleness
- Rethinking access control for RAG with Amazon Quick and Amazon Bedrock
- How Qlik built grounded, enterprise-scale AI with Amazon Bedrock
- How to Build a Multi-Modal RAG Pipeline for Images and Documents in Python
Coding & dev tools
AI-assisted coding and Python tooling · 6
- Python 3.15 Has a New Profiler
Python 3.15 introduces a new built‑in profiler that captures performance details missed by the traditional cProfile module, offering developers deeper insight into execution bottlenecks.
- Cline Bot makes the case for an open AI coding harness
Cline Bot Inc. argues for an open AI coding harness to bridge open-weight models with developer workflows, emphasizing the trade‑off between cost and capability.
- Meta’s Muse Spark Open now available in Amazon SageMaker JumpStart
Meta’s Muse Spark Open, a coding‑optimized open‑weight large language model, has been added to Amazon SageMaker JumpStart, letting developers spin up an AI endpoint directly in their AWS account via SageMaker Studio or the SageMaker…
- How to Fine-Tune Llama 3 for Custom Tool Calling with Unsloth in Python
The article demonstrates how to fine‑tune Meta’s Llama 3 model with the Unsloth library to reliably generate structured JSON for custom API tool calls in Python.
- Vital Skills: Upskilling Genie Code with Healthcare and Life Sciences skills
Genie Code is a unified, governed AI coding agent that has been upskilled with Healthcare and Life Sciences domain expertise, enabling developers to generate code that adheres to sector‑specific regulations and data standards.
- The Pulse: new trend of building “internal vibe-coding apps” at tech companies
Ramp and Stripe are rolling out internal “vibe‑coding” platforms—Glass and Harbor—that let non‑engineers rapidly prototype complex internal applications using AI.
Cloud platforms & enterprise
Bedrock, AWS and enterprise adoption · 39
- Access OpenAI GPT models on Amazon Bedrock from Canada with cross-Region inference
Amazon Bedrock now enables Canadian customers to invoke OpenAI GPT‑5.6, GPT‑6, and GPT‑6.1 models from Canada (Central) and Canada West (Calgary) via cross‑Region inference, routing requests to US regions while preserving North‑America…
- Fix evaluation bias to unlock better models on Amazon Bedrock
Property Finder discovered that 96.8 % of their evaluation labels were inherited from a single incumbent vision model on Amazon Bedrock, causing a systematic bias that made every alternative model appear inferior.
- Cloudera bets portability, not agents, will set it apart in enterprise AI
Cloudera is positioning portability of AI workloads as its differentiator for enterprise customers, emphasizing control over model and data execution environments amid regulatory pressure, token cost concerns, and lock‑in fears.
- Deliberate decisions and enterprise AI value
The article discusses the accelerating pace of AI progress and rising expectations within enterprises, emphasizing the need for deliberate decision‑making to capture AI value.
- Kore.ai launches Autoloop to keep tuning enterprise AI agents after they go live
Kore.ai’s new Autoloop is an optimization engine that hooks into its Kore.ai Agent Platform to automatically fine‑tune AI agents after they go live.
- MIT affiliates named 2026 Schmidt Sciences AI2050 Fellows
MIT affiliates, including five faculty members and two alumni, have been named to the 2026 cohort of Schmidt Sciences’ AI2050 Fellows, a philanthropic program focused on solving hard problems in artificial intelligence.
Show 33 more in Cloud platforms & enterprise
- How Eon built multi-cloud investigation with an MCP Gateway on Amazon Bedrock AgentCore
- Deploy AI in a HIPAA-eligible environment with Amazon Quick
- How Bayer Crop Science built an agentic decision engine on Amazon Bedrock AgentCore with AWS Unified Operations
- Configure the ChatGPT desktop app with Amazon Bedrock in AWS GovCloud (US)
- How Octave built a secure multi-tenant AI foundation on Amazon Bedrock
- How IskayPet unified corporate knowledge with Amazon Bedrock Managed Knowledge Bases
- Structured outputs for Anthropic Claude models on Amazon Bedrock
- Visualize points and routes with GeoJSON in Amazon Quick Sight layer maps
- Pegasus 1.5 from TwelveLabs is now available in Amazon Bedrock
- Accessing OpenAI GPT models on Amazon Bedrock with global cross-Region inference in Thailand, Malaysia, Singapore, Indonesia, and Taiwan
- Pay-per-inference for AI agents: How BlockRun and Incarna use Amazon Bedrock AgentCore payments
- Building ambient agents with Amazon Bedrock AgentCore: From event-driven signals to human-in-the-loop workflows
- How Rescale built an AI triage agent with Strands Agents and AI-DLC
- Build long-horizon agents with Grok 4.7 on Amazon Bedrock
- Generate presentations with Amazon Bedrock AgentCore
- Introducing GLM 5.3 on Amazon Bedrock
- IBM puts governed context at the center of production AI
- Building agents that transact with Anchor Browser and AgentCore payments
- How ShipStation Global built an AI IT help desk on Amazon Bedrock AgentCore
- How Postman runs Agent Mode for 40 million developers on Amazon Bedrock
- Automate remediation post AWS DevOps Agent investigation
- Natural language security queries at Orca Security using Amazon Bedrock
- ICYMI: What landed for AI builders in September 2026
- Introducing Claude Haiku 5.5 on AWS
- How U.S. Hunger connects leaders and data using Amazon Quick
- Google, Anthropic and others to support new AI crawling opt-out standard
- Managing Amazon Bedrock costs end to end: A playbook for FinOps and platform teams
- Amazon Polly delivers a major upgrade to its Generative engine
- Manage private pricing offers for Amazon Bedrock with managed entitlements
- McKinsey connects enterprise data through a knowledge graph for AI
- How Cornerstone OnDemand cut database diagnosis by 78% with Amazon Bedrock
- Responsible AI governance: How AWS positions customers to align with ISO/IEC 42005:2025
- What to expect during monday.com’s Elevate event: Join theCUBE Oct. 29
Everything else
Research and news that fits no single group · 1
- 4 insights from Smarsh EDGE: Governance becomes the foundation for communications intelligence
The Smarsh EDGE event highlighted a shift from treating employee communications solely as regulatory archives to leveraging them as a source of business intelligence through artificial intelligence.
Grouped by each post's main topic tag. Summaries are AI-generated from the source and always link to it.
Get next week's releases by email
No hype. The week's best AI engineering posts, every Sunday.