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
Lyft’s Feature Store: Architecture, Optimization, and Evolution
Lyft built a centralized Feature Store to optimize feature management, deployment, and data consistency for machine learning workflows across its rideshare platform. The system ingests batch features defined by Spark SQL and JSON metadata into Airflow DAGs, which simultaneously write data to Hive for model training and to an online serving layer for low-latency inference. Real-time streaming features are processed with Apache Flink and Apache Kafka before being ingested into the serving infrastructure. The online layer, dsfeatures, couples DynamoDB with a ValKey write-through LRU cache and OpenSearch to serve embeddings. Infrastructure optimizations reduced standard P95 read latency by a third while supporting over a trillion additional read and write operations.
Dynamic context discovery
Cursor describes dynamic context discovery, a coding-agent pattern that provides fewer details initially and lets the agent retrieve relevant information instead of injecting everything as static context. The approach uses files as a simple interface: long tool responses, chat history, Agent Skills, MCP tool descriptions, and integrated terminal output are stored or synchronized so the agent can inspect only what it needs with tools such as tail, grep, rg, and jq. For MCP, only tool names are included initially, while descriptions are synced into one folder per server; an A/B test found a 46.9% reduction in agent tokens for runs that called an MCP tool, with high variance based on the number of MCPs. Cursor says the design reduces context bloat, preserves information truncation could lose, supports recovery after summarization, and can expose MCP re-authentication status. The changes are planned to become available to all users in the coming weeks.
NVIDIA Cosmos Reason 2 Brings Advanced Reasoning To Physical AI
NVIDIA released Cosmos Reason 2, an open reasoning vision-language model designed to improve multi-step planning, uncertainty handling, and physical AI decision-making. Available in 2B and 8B parameter configurations, the model increases input context support to 256K tokens, up from 16K in the prior version. The release adds spatial perception capabilities including optical character recognition, 2D and 3D point localization, bounding boxes, and trajectory generation for robotics. In domain adaptation tests on autonomous vehicle video captioning, fine-tuning the 8B model improved BLEU scores by 10.6% and LingoQA scores by 13.8%. The model leads the Physical AI Bench and Physical Reasoning leaderboards for open visual understanding models.
New research: Customer service team evolution
An analysis of 166 interviews examines how customer service roles, workflows, and team structures changed after adopting Fin or similar AI agents. Using interviews with Intercom customers or prospects, the researchers cleaned conversation transcripts with Python and coded them through the OpenAI API against a closed codebook. Workflow disruption was reported by 94.58% of participants, while 82.53% reported changed responsibilities; 27.71% reported headcount or hiring effects, and 6.02% reported changed team structures or reporting lines. The findings describe hybrid human–AI operations in which AI handles triage, routing, translation, repetitive responses, and some refunds, while people increasingly oversee quality, testing, metrics, and complex or sensitive interactions, although organizational strategies and required skills remain uneven.
Introducing Falcon-H1-Arabic: Pushing the Boundaries of Arabic Language AI with Hybrid Architecture
The Technology Innovation Institute introduced Falcon-H1-Arabic, a family of Arabic language models released in 3B, 7B, and 34B parameter sizes. Built on a hybrid architecture, each block runs Mamba State Space Models and Transformer attention in parallel before fusing their representations at the output projection. Pre-training incorporated around 300 billion tokens across Arabic dialects, English, and multilingual content filtered via linguistic analysis. A post-training pipeline combining supervised fine-tuning and direct preference optimization addresses long-context utilization and alignment. Across evaluations on the Open Arabic LLM Leaderboard, the 34B model achieves roughly 75%, exceeding the performance of larger baseline systems like Llama-3.3-70B.
NVIDIA brings agents to life with DGX Spark and Reachy Mini
At CES 2026, NVIDIA presents a desk-based personal assistant that combines DGX Spark processing with a Reachy Mini robot, allowing an agent to converse, perceive camera input, speak, and perform actions. The tutorial assembles NVIDIA Nemotron 3 Nano for reasoning, NVIDIA Nemotron Nano 2 VL for vision, ElevenLabs for text-to-speech, and Reachy Mini hardware or simulation through the NeMo Agent Toolkit and Pipecat. Its configuration routes text, visual, and action requests to different models or a ReAct tool-calling agent, while the robot daemon exposes movement and state capabilities. The guide covers local, cloud, and serverless deployment, then starts the services needed for voice and vision interaction; the completed setup greets users and supports example text and camera prompts.
How Merge brought together all it knows using Notion AI
Merge found that company knowledge was scattered across Slack, Google Docs, Salesforce, and other tools, causing repeated questions, slow onboarding, and time lost searching for context. It used Notion AI as a shared knowledge layer, connected Salesforce and Slack, and gave teams guidance on what content to add, including detailed project documents, meeting transcripts, and structured databases. Teams populated the workspace with content calendars, sales enablement materials, hiring plans, roadmaps, and an AI Prompt Library, while all-hands and team meetings demonstrated practical workflows. Merge reports that employees now ask Notion AI first, saving each team member several hours per week, helping teams make decisions with less manual searching, and making faster information retrieval one of the most-cited contributors to improved satisfaction in its latest company-wide survey.
AprielGuard: A Guardrail for Safety and Adversarial Robustness in Modern LLM Systems
Modern language model systems face security and safety threats spanning prompt injection, jailbreaks, and tool manipulation across long-context agentic workflows. To address these vulnerabilities, researchers introduced AprielGuard, an 8B parameter causal decoder-only transformer based on a downscaled Apriel-1.5 Thinker Base model. The model detects 16 safety risk categories alongside binary adversarial attacks across standalone prompts, multi-turn dialogues, and agentic workflows. Trained on synthetic datasets created with Mixtral-8x7B, NVIDIA NeMo Curator, and SyGra, AprielGuard operates in both explainable reasoning and fast non-reasoning modes. Evaluation shows high precision and recall on safety and adversarial benchmarks, though trade-offs exist between explainability latency and non-reasoning performance.
Demystifying user journeys: Revolutionizing troubleshooting with auto tracking
Troubleshooting critical issues in the Grab app was hindered by incomplete user-journey data, because manually instrumenting hundreds of UI components left interaction gaps and delayed diagnosis. AutoTrack is a new SDK that records application state, user interactions, and UI screens in real time, creating an end-to-end session chronicle from launch through navigation. It tracks lifecycle transitions and launch mechanisms such as deeplinks and push notifications, while UIViewController, Activity, and Fragment lifecycle events identify the displayed screen. Screen metadata includes the full hierarchy as JSON; after personal information is redacted for GDPR/PDPA compliance, that data is used to train an AI model that generates an HTML representation of the screen and interactions. AutoTrack supports reproducing elusive bugs, automated issue assignment, UI test generation, and analytics-trigger validation, with future work adding network and performance signals, privacy guardrails, and automated triage.
Steam, Steel, and Infinite Minds
The post frames AI as a new “miracle material” that could reshape knowledge work as steel and semiconductors reshaped earlier eras. It argues that coding agents already let one programmer orchestrate several AI coding agents, but broader knowledge work remains constrained by fragmented context across tools and weak verifiability. At the organisational level, it compares AI with steel and steam, suggesting agents could preserve workflow context, surface decisions, and reduce dependence on meetings and hierarchical approvals; Notion is described as using more than 700 agents alongside 1,000 employees for repetitive tasks. At economy scale, it predicts organisations spanning humans and thousands of agents, with continuous workflows and new rhythms, while acknowledging that this transition may be disorienting and requires redesign rather than simply adding chatbots to existing tools.
First Block – Tools & craft – Notion
First Block is a set of conversations with founders and company leaders across legal technology, finance, career access, software development, talent, go-to-market, and developer tools. The episodes feature Gabe Pereyra of Harvey, Gerry Giacomán Colyer of Clara, Garrett Lord of Handshake, Jesse Zhang of Decagon, Fabian Hedin of Lovable, Brendan Foody of Mercor, Varun Anand of Clay, and Michael Truell of Cursor. Topics include deploying AI agents in finance operations across Latin America, democratizing software development and career opportunities, intentional customer discovery, and building for current AI capabilities. The collection also covers founder-led selling, product-market fit, specialized go-to-market platforms, commercial thinking, startup pivots, and departures from conventional startup advice.
That’s a Wrap: MongoDB’s 2025 in Review & 2026 Predictions
MongoDB recapped its product releases, executive transitions, and enterprise adoption milestones throughout 2025 alongside predictions for 2026. The company acquired embedding and reranking specialist Voyage AI to mitigate large language model hallucinations and launched MongoDB AMP to accelerate legacy application modernization. Vector search and full-text search capabilities were expanded from MongoDB Atlas to Community Edition and Enterprise Server, enabling on-premises and hybrid retrieval-augmented generation workloads. Enterprise implementations included pharmaceutical distributor McKesson scaling tracking infrastructure to handle 1.2 billion containers annually after replacing its monolithic legacy system. Executive predictions for 2026 emphasized human-in-the-loop verification, grounding language models with factual retrieval, and building multi-cloud resilience against outages and data sovereignty constraints.
Hooks for security and platform teams
Cursor has announced ecosystem partnerships around hooks, which run before or after defined stages of its agent loop to observe, block, or modify behavior. The integrations target MCP governance, code security, dependency scanning, agent safety, and secrets management, connecting Cursor with security tooling, observability platforms, secrets managers, and compliance systems. MintMCP, Oasis Security, and Runlayer provide MCP inventory, least-privilege enforcement, audit trails, and centralized tool-interaction visibility, while Corridor and Semgrep give coding-time feedback and vulnerability scanning. Endor Labs intercepts package installations to scan for malicious dependencies, Snyk’s Evo Agent Guard reviews agent actions for prompt injection and dangerous tool calls, and 1Password validates that all required environment files from 1Password Environments are properly mounted before shell commands execute, enabling just-in-time secrets access without writing credentials to disk.
Closing the Books Without the Spreadsheet Shuffle: My Fall Internship at Ramp
During a fall internship at Ramp, the author worked on reducing repetitive accounting work around month-end close, focusing on amortization and ERP mapping workflows. One project uses receipt data, merchant category codes, customer history, and embedding similarity to detect likely prepaid expenses and suggest service periods and accounts or templates, with recommendations reviewable before application. A second built Ramp-Native Amortization to create schedules, generate periodic debit and credit journal entries, handle catch-ups and reclassifications, and sync entries to ERPs through a dashboard; transaction and reimbursement support was nearing alpha production release, while bill pay was next. The third generates high-confidence mapping-rule suggestions from customer coding patterns on a nightly schedule. Across the work, backtesting, accuracy metrics, and human workflow research support automation intended to earn trust rather than remove accountant judgment.
Token-count-based Batching: Faster, Cheaper Embedding Inference for Queries
Serving embedding models for short search queries often suffers from poor GPU efficiency because traffic is spiky and memory-bound. Traditional time-window and request-count batching strategies lead to inconsistent GPU utilization, while tensor padding wastes compute on empty tokens. To resolve these bottlenecks, Voyage AI implemented token-count-based batching paired with padding removal in inference engines like vLLM. The architecture uses Redis with Lua scripts to atomically aggregate pending requests until reaching an optimal hardware saturation token threshold. Across production deployments, this approach achieved a 50% reduction in GPU inference latency with 3X fewer GPUs and improved throughput by up to 8×.
Make with Notion Showcase Europe 2025: Highlights & Takeaways
Make with Notion Showcase Europe 2025 brought more than 1,300 attendees across four cities during the second year of the in-person conference series. The event centered on three themes: agentic AI taking action, organized knowledge enabling effective AI, and a community building AI-powered work. Notion presented Notion 3.0, whose Agent can handle multi-step workflows, and previewed Custom Agents that can be shared across teams, scoped to databases or teamspaces, and triggered by schedules or events. With MCP integration, Notion content and connected app context can be accessed from other tools; Custom Agents are in early access and planned for broader availability in early 2026. Research cited losses of up to €16,000 per employee annually in the UK, France, and Germany from searching for scattered information, while customer sessions covered knowledge management, AI workflows, and Notion implementations.
How Grab is accelerating growth with real-time personalization using Customer Data Platform scenarios
Grab’s Customer Data Platform (CDP) introduced Scenarios to address the limits of daily batch attributes for real-time personalization across marketing and user journeys. The capability combines Scribe event triggers, geo-fencing, optional historical CDP data, and predictive models, with a self-serve GUI for configuring and validating event sequences and APIs for programmatic management. Incoming events are processed through Flink, with historical data evaluated in StarRocks and outputs written to Kafka, Amphawa, or both for downstream consumption; configured Scenarios become available within an hour and are designed for latency under 15 seconds. More than 12 implementations are live, including a Grab Unlimited abandonment intervention that notified users 15 minutes after signup drop-off and produced over a 3% uplift in subscriber conversions versus non-real-time acquisition campaigns. Planned work includes scalable Scenario sharding, multiple output destinations, and delayed triggering.
Tokenization in Transformers v5: Simpler, Clearer, and More Modular
Transformers v5 overhauls its tokenization framework by separating tokenizer architecture from trained vocabularies. In contrast to v4's dual slow Python and fast Rust files, v5 consolidates each model tokenizer into a single file defaulting to the Rust-backed TokenizersBackend. The pipeline stages—normalizer, pre-tokenizer, model algorithm such as BPE or Unigram, post-processor, and decoder—are now directly exposed and configurable rather than buried in serialized files. Practitioners can instantiate blank tokenizer architectures and train custom vocabularies directly from iterators using native methods like train_new_from_iterator while retaining model-specific formatting rules. The wrapper layer continues to bridge raw tokenization and model requirements by managing chat templates, context limits, and special token insertion.
First Block with Garrett Lord, Co-founder and CEO of Handshake
First Block features Garrett Lord, co-founder and CEO of Handshake, discussing the company’s path from his early rejection by hundreds of universities and periods living out of a Ford Focus to building a career platform serving 17 million users and nearly 1 million companies. He describes Handshake’s culture as “Olympic pace,” arguing that extreme clarity about standards is preferable to consensus because it helps the right people self-select and avoids diluting what makes the company special. The conversation also covers scaling through leadership, Handshake AI, and the future workforce. Lord’s central advice is to stack daily effort even when operating at 60% and to go all in on major opportunities, particularly as AI reshapes the economy.
The Open Evaluation Standard: Benchmarking NVIDIA Nemotron 3 Nano with NeMo Evaluator
Assessing whether large language model improvements stem from genuine advances or underspecified evaluation conditions remains a major challenge across the industry. Most published model evaluations omit critical execution parameters, prompt templates, harness versions, and runtime configurations. In response, NVIDIA released Nemotron 3 Nano 30B A3B alongside its complete, reproducible evaluation recipe built with the open-source NeMo Evaluator library. The library serves as an orchestration layer that unifies diverse benchmark harnesses under standard configurations while decoupling evaluation logic from underlying inference backends. Developers can execute the identical evaluation pipeline against local deployments or hosted endpoints using published YAML configurations and structured logging.
Top topics
- Architecture834
- Performance816
- LLMs660
- Machine Learning622
- Open Source563
- AI Agents553
- AI470
- Developer Experience448
Weekly digest
The best engineering reads, once a week.