---
title: "Latest reads"
description: "The engineering internet, summarised so you can actually read it."
---

# Latest reads
> The engineering internet, summarised so you can actually read it.

## Articles

### [Code-Switching Detection via Language-Selective Dimensions in Speech Foundation Models](https://yomu.fyi/post/code-switching-detection-via-language-selective-dimensions-in-speech-f.md)
- Company: [Apple](https://yomu.fyi/company/apple.md)
- Published: Oct 14, 2026
- Type: Explainer

Code-switching, the alternation between languages in a single utterance, is common in multilingual speech but underrepresented in training corpora for speech models. To address this, a lightweight method detects code-switching directly from speech without requiring multilingual transcriptions. The approach leverages language-selective dimensions in pretrained speech encoders by identifying hidden-state dimensions selective to each language from monolingual recordings and aggregating frame-level activations to predict code-switching. On the CS-FLEURS dataset covering 14 language pairs, the Whisper-small encoder achieves a mean F1 of 0.798, outperforming a CNN baseline (0.776) and a HuBERT linear baseline (0.751). These results show that pretrained speech encoders inherently encode the information necessary for code-switching detection, offering a scalable tool for building more representative multilingual corpora.


### [Modal crosses $1B in Annualized Run Rate and raises $750M Series D](https://yomu.fyi/post/modal-crosses-1b-in-annualized-run-rate-and-raises-750m-series-d.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Author: Erik Bernhardsson
- Published: Oct 14, 2026
- Type: Announcement

Modal crossed $1B in annualized run rate and raised $750M in Series D funding at a $15.75B post-money valuation, led by Accel. Usage has tripled since the Series C five months ago, with year-over-year growth exceeding 15x, and the company now powers applications from companies including DoorDash, Cognition, and Runway. To avoid the friction of the old stack, Modal built its own file system, container runtime, scheduler, and networking from scratch on a capacity pool spanning over 20 clouds. Sandboxes, launched in 2023 for running LLM-generated code, now represent a $300M run rate product with nearly 3 billion sandboxes spun up and scaling to a million concurrent. New products include multi-node training with a single decorator giving a fully RDMA-connected cluster, Modal servers for HTTP serving, Endpoint Candidates for LLM serving, and generally available virtual machines.


### [One SDK, Multiple Platforms: Introducing the Auth0 SDK for Kotlin Multiplatform](https://yomu.fyi/post/one-sdk-multiple-platforms-introducing-the-auth0-sdk-for-kotlin-multip.md)
- Company: [Auth0](https://yomu.fyi/company/auth0.md)
- Author: Prince Mathew
- Published: Oct 14, 2026
- Type: Announcement

Auth0 released its official Kotlin Multiplatform SDK for Android and iOS, enabling unified authentication logic in shared common code. Developers write login, logout, token management, and direct API calls once in commonMain using Kotlin suspend functions and a sealed Result type for exhaustive error handling. Native security is maintained through platform-specific implementations: Android uses Keystore-backed DataStore and Custom Tabs, iOS uses Keychain and ASWebAuthenticationSession, with Swift async/await interop via SKIE. The SDK is available on Maven Central as com.auth0.kmp:auth0:1.0.0-beta.1 as a single artifact or separate modules, requiring Kotlin 2.3.21, Android minSdk 24, and iOS 14.0+. It includes a Credentials Manager for token storage and automatic renewal, and supports advanced features including WebAuthn passkeys, DPoP token binding, custom storage, and configurable retry with fixed or exponential backoff.


### [How to Retrieve a Phone Number's SMS Pumping Score in Rust](https://yomu.fyi/post/how-to-retrieve-a-phone-number-s-sms-pumping-score-in-rust.md)
- Company: [Twilio](https://yomu.fyi/company/twilio.md)
- Author: Matthew Setter
- Published: Oct 14, 2026
- Type: Tutorial

SMS Pumping Fraud exploits phone number input fields to generate revenue from SMS messages sent to numbers controlled by a specific mobile network operator. Twilio’s Lookup v2 API provides an SMS Pumping Risk Score from 0 to 100, based on a proprietary model using network data including Verify Fraud Guard signals and unusual traffic patterns. This Rust tutorial uses the rustlio crate, dotenvy, serde, and tokio to call the API and retrieve the score along with carrier risk category and block status. Canadian numbers do not return carrier information, and Twilio advises against using the score for US or Canadian numbers. Guidelines recommend sending messages for scores 0-60, adding friction for 60-75, treating as suspicious for 75-90, and blocking for 90-100.


### [How to Rewrite Apache Spark UDFs for GPU Acceleration with AI Agents](https://yomu.fyi/post/how-to-rewrite-apache-spark-udfs-for-gpu-acceleration-with-ai-agents.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Elizabeth Goodman
- Published: Oct 13, 2026
- Type: Tutorial

User-defined functions (UDFs) in Apache Spark break the declarative abstraction, forcing GPU-accelerated pipelines to fall back to CPU execution and incur data transfer overhead. NVIDIA cuDF addresses this with GPU parallelism, but UDFs remain a bottleneck due to their opaque code. The post describes three approaches to rewrite UDFs into GPU-accelerable forms: SQL expressions, the RapidsUDF interface using cuDF columnar operations, or custom CUDA kernels via JNI. The cuDF skills for Apache Spark automate this process with four agent-guided feedback loops for testing, conversion, benchmarking, and optimization, providing a project template and best practices. A production deployment converted over 30 UDFs, achieving a 22x speedup on a 600-billion-row workload, reducing operator time from 274.55 to 12.78 hours.


### [Scope workloads and share resilience policies with AWS Resilience Hub](https://yomu.fyi/post/scope-workloads-and-share-resilience-policies-with-aws-resilience-hub.md)
- Company: [AWS](https://yomu.fyi/company/aws.md)
- Author: Sinisa Mikasinovic
- Published: Oct 13, 2026
- Type: Tutorial

AWS Resilience Hub’s Next generation, announced September 18, 2026, adds three capabilities to align workload scope and resilience requirements with running environments. Amazon EKS label support selects Kubernetes resources using existing labels with AND logic, establishing precise resource boundaries without reorganizing namespaces. Generative AI-powered Dependency Insights summarizes discovered dependencies and highlights patterns warranting resilience review, such as cross-Region dependencies questioning recovery design. Resilience policy sharing through AWS Organizations lets a central team define requirements once and make them available to member accounts, with policy events tracking adoption. The post walks through scoping a checkout service on EKS using labels, examining dependencies via synthetic traffic, and applying a shared tier-1-critical policy with a 99.9% availability SLO, 5-minute multi-AZ RTO, 1-minute RPO, 15-minute multi-Region RTO, and 60-minute backup intervals.


### [Databricks is a Visionary the Gartner® Magic Quadrant™ for Analytics and Business Intelligence Platforms](https://yomu.fyi/post/databricks-is-a-visionary-the-gartner-r-magic-quadranttm-for-analytics.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Chao Cai, Kayli Berlin, Miranda Luna
- Published: Oct 13, 2026
- Type: Announcement

Databricks debuted as a Visionary in the 2026 Gartner Magic Quadrant for Analytics and Business Intelligence Platforms, achieving the highest debut performance in the report’s 20+ year history. The company also received one of the highest Completeness of Vision scores among all vendors evaluated. Databricks attributes this recognition to its ability to deliver on three major trends reshaping businesses: data and AI platform convergence, agentic analytics, and semantics at the core. The Databricks lakehouse integrates BI directly, eliminating data silos and the need for extracts, so agents can reason over complete, governed context. With Genie and AI/BI, the platform automates complex workflows and provides a unified semantic layer through Unity Catalog Business Semantics and Genie Ontology.


### [What's New in Lakeflow Designer - October 2026 Roundup](https://yomu.fyi/post/what-s-new-in-lakeflow-designer-october-2026-roundup.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Jason Messer, Emanuel Zgraggen, V Maharajh, Kyrylo Bakumenko, Xiao Yan, Matt Jones
- Published: Oct 13, 2026
- Type: Announcement

Lakeflow Designer, a visual no-code data preparation tool in Databricks, added three new operators for cleaning, deduplication, and quality checks, addressing customer requests for more no-code capabilities. The Prepare operator standardizes data by changing case, converting types, replacing values, and applying formulas; Unique removes duplicates across selected columns with ordering control; Guardrails enforces quality rules with configurable failure handling. Existing operators gained multi-table joins with separate outputs for matched, left unmatched, and right unmatched rows, branching in Filter, and dynamic column selection in Select. Output options expanded to materialized views in Unity Catalog that refresh automatically, merge into existing tables, and direct file writes to CSV, Excel, or JSON with dynamic worksheets. Canvas improvements include a redesigned interface with better contrast and larger connection targets for large workflows, a searchable table of contents, visual version history, and live cursor presence.


### [Run fast, private agents on-device with LFM2.5-24B-A2B](https://yomu.fyi/post/run-fast-private-agents-on-device-with-lfm2-5-24b-a2b.md)
- Company: [Hugging Face](https://yomu.fyi/company/hugging-face.md)
- Author: Raahul Vignesh, Song Duong, Akhilesh Deepak Gotmare, Fernando Fernandes Neto, Jiahui Wang, Mayank Agarwal, nathan ranchin, Sergei Tilga, Sinoué GAD, Tim Seyde, Zetian Li, changyi yang, Simon Lee, Leonie Monigatti, Maxime Labonne, Karthik Macherla, Alfred Eng, Khoi Nguyen, Paul Pak, xx, Tarek Dakhran, Anthony Maltsev
- Published: Oct 13, 2026
- Type: Announcement

LFM2.5-24B-A2B is a 24-billion-parameter reasoning model with only 2 billion active parameters per token, designed for fast, private on-device agents. It uses a Mixture-of-Experts architecture with 40 layers, 64 experts per block, and top-4 routing, combined with gated short convolutions and GQA. The model produces an explicit chain of thought before answering and supports tool calling and multi-step tasks. Pre-training scaled to 33 trillion tokens, followed by mid-training and a five-stage post-training pipeline including supervised fine-tuning, multi-domain distillation, and agentic reinforcement learning. On benchmarks, it leads or matches larger models on tool use and instruction following, achieving 48.09 on LiveClaw and 96.75 on IFEval, while decoding 184 tokens per second on an Apple M5 Max.


### [Why Your Sandbox Never Quite Matches Production (And Why That's Fine)](https://yomu.fyi/post/why-your-sandbox-never-quite-matches-production-and-why-that-s-fine.md)
- Company: [Stack Overflow](https://yomu.fyi/company/stack-overflow.md)
- Author: Sancharini Panda
- Published: Oct 13, 2026
- Type: Explainer

The gap between sandbox and production environments is structural and inevitable because they change on different schedules with different triggers. Infrastructure divergence rarely affects what integration tests catch, but behavioral contract drift—when simulated upstream responses no longer match real ones—causes tests to pass while production fails. The most effective teams connect simulation updates to upstream deployment events rather than to test failures, using observation-based tools like Keploy that capture real HTTP traffic to keep simulations current. Full production parity is not the goal; targeted accuracy for the specific behavioral contracts that integration tests validate is sufficient to catch the failures that matter most in distributed systems. The sandbox covers what it can, production reveals what it cannot, and the maintenance work is keeping the sandbox’s coverage accurate within its boundary, not expanding it to match production exactly.


### [Redesigning source control at Anthropic](https://yomu.fyi/post/redesigning-source-control-at-anthropic.md)
- Company: [Anthropic](https://yomu.fyi/company/anthropic.md)
- Author: Anthropic
- Published: Oct 13, 2026
- Type: Explainer

At Anthropic, Claude’s ability to iterate on code at high speed has caused git-based source control to become the bottleneck in their development workflow. To handle the growth from 2,500 to over 30,000 daily code merges, the team built Scion, an in-house version control system, fully migrating their monorepo from git in five months. Scion diverges from git by assuming clients are always online, shifting computation to the server, and using horizontally scalable data structures without any fully materialized copy of the repository. The system stores changes as entries in a write-ahead log, supports sparse and virtualized checkouts, and enables cheap, lock-free commits that scale with change size rather than repository size. Anthropic shares Scion’s design principles to help the broader ecosystem prepare for AI-driven scaling challenges, noting that similar pressure points will emerge for other organizations as AI coding tools become more common.


### [Under the hood, off the hot path: How D1 scales to millions of databases](https://yomu.fyi/post/under-the-hood-off-the-hot-path-how-d1-scales-to-millions-of-databases.md)
- Company: [Cloudflare](https://yomu.fyi/company/cloudflare.md)
- Author: Lambros Petrou
- Published: Oct 13, 2026
- Type: Explainer

D1, Cloudflare’s serverless SQLite database, has seen 15x growth in live databases over the past year and scales by keeping query execution on the edge in Durable Objects, separate from the control plane’s metadata operations. The control plane uses an Edge API Worker with Hyperdrive to access Postgres clusters for transactional metadata, while Workers KV caches database-to-Durable-Object mappings for fast edge lookups. Database creation completes in about 1.6 seconds by generating a Durable Object ID, persisting metadata to Postgres, and storing the mapping in Workers KV, with the Durable Object lazily initialized on first query. Storage usage enforcement uses atomic counters in Workers KV that are checked locally on each query and updated asynchronously after execution, avoiding any central database round trip during the query. Precomputed aggregations for billing are stored in ClickHouse, and guiding principles such as eventual consistency and idempotency further reduce reliance on centralized components.


### [Making password login safer on Notion](https://yomu.fyi/post/making-password-login-safer-on-notion.md)
- Company: [Notion](https://yomu.fyi/company/notion.md)
- Author: Ankit Gupta
- Published: Oct 13, 2026
- Type: Announcement

Notion is requiring a second step of verification for password login to protect against credential reuse from other site breaches, and this change will roll out in stages over the next few months. When users set or change a password, they will be guided to set up two-step verification via an authenticator app or phone number and should save the backup codes provided. Users who already have two-step verification on their password need no action, and those without a password can continue using passkeys, email codes, or SSO. For password-only users without two-step verification, Notion will provide notice and time to set up a second step before turning off password-only login; if they choose not to, another method like an email code can still grant access. Additionally, where Notion can verify that another working method exists, it plans to turn off password login for those unused passwords.


### [Meta’s Generative Ads Model (GEM): Scaling the Central Model and Compute Engine for Frontier Ads AI](https://yomu.fyi/post/meta-s-generative-ads-model-gem-scaling-the-central-model-and-compute.md)
- Company: [Meta](https://yomu.fyi/company/meta.md)
- Author: Huayu Li
- Published: Oct 13, 2026
- Type: Announcement

Meta introduced GEM, the industry’s largest recommendation system foundation model, and its 2026 generation shifts from isolated per-domain optimizations on Facebook, Instagram, and Reels to a single central foundation model. The Kunlun architecture, co-designed with the latest GPU hardware, enables 2.5x model complexity and demonstrates LLM-like scaling laws with block-aligned sparse attention reducing sequence complexity. The LoopFM post-training technique transfers GEM’s internal representations to downstream models, doubling intelligence transfer at no additional compute cost or latency. GEM delivered 18.9% conversion improvement and 12% ad click-through rate improvement across all Meta ad surfaces and funnel stages, including 5.8% conversion uplift on Facebook, 4.1% on Instagram, and 1.6% on off-site conversion. Training efficiency doubled to 20-25% model FLOPS utilization while scaling total training FLOPs by 4x across thousands of GPUs using custom kernels and 5D parallelism.


### [Rewinding the exponential: How we keep hardware usage down](https://yomu.fyi/post/rewinding-the-exponential-how-we-keep-hardware-usage-down.md)
- Company: [Cloudflare](https://yomu.fyi/company/cloudflare.md)
- Author: Christian Endres
- Published: Oct 13, 2026
- Type: Problem & solution

Cloudflare’s Data team manages a pipeline handling over two billion events per second from tens of thousands of servers, powering Logs, SQL API, and observability. To counter exponential traffic growth against linear hardware expansion, they implemented several optimizations over five years, deferring two and a half years of hardware expansion. In 2022, they replaced a monolithic schema with product-specific schemas, reducing backbone bandwidth by 35-40%. In 2024, they switched compression from LZ4 to zstd, achieving a 40% reduction in compressed traffic with negligible CPU impact. In 2026, they decoupled pipeline transit from database redundancy via a shared ingest pool and late replication, applied sampling downstream, and introduced aggregations, eliminating about 70 GB/s of compressed traffic and halving projected 2031 data volume.


### [Postgres at Open House NYC](https://yomu.fyi/post/postgres-at-open-house-nyc.md)
- Company: [Clickhouse](https://yomu.fyi/company/clickhouse.md)
- Published: Oct 13, 2026
- Type: Announcement

At ClickHouse Open House NYC, the team announced WalShadow, an open-source engine that replicates Postgres into ClickHouse from the physical WAL, achieving sub-second latency and high throughput. They also launched ClickHouse Managed Postgres on Google Cloud in private preview, offering Postgres on local NVMe storage with native CDC into ClickHouse. The chdb Postgres extension was released, enabling fast imports from cloud storage using the embeddable ClickHouse engine. A live demo showed WalShadow keeping ClickHouse under 400 ms behind during a simulated trading platform with 50,000 trades per second, and detecting a fraud burst in under 500 ms compared to over 15 seconds with ClickPipes. The integrated stack aims to provide world-class Postgres for transactions and ClickHouse for analytics, with WalShadow as a key step toward unifying OLTP and OLAP.


### [Behind the messenger: making messaging human - The Intercom Blog](https://yomu.fyi/post/behind-the-messenger-making-messaging-human-the-intercom-blog.md)
- Company: [Intercom](https://yomu.fyi/company/intercom.md)
- Published: Oct 13, 2026
- Type: Explainer

Intercom redesigned its messenger to make business communication personal, moving away from the anonymity of the early internet. The post argues that sharing real names, faces, and job titles builds trust, as user testing showed customers repeating the names of the reps they messaged. System-generated text like active status, location, and local time that cannot be changed adds authenticity and helps users understand the asynchronous nature of the system. The post recounts an incident where a customer doubted a rep’s real name, highlighting the need for transparency, and notes that job titles prevent white lies like claiming to pass feedback to a product team when the rep is on that team. The post concludes that these small details together build trust, leading to customer loyalty and forgiveness when things go wrong, and that users initially don’t expect this information but quickly come to expect it everywhere.


### [Basecamp's Jason Fried on product strategy - The Intercom Blog](https://yomu.fyi/post/basecamp-x27-s-jason-fried-on-product-strategy-the-intercom-blog.md)
- Company: [Intercom](https://yomu.fyi/company/intercom.md)
- Published: Oct 13, 2026
- Type: Explainer

Jason Fried, co-founder of Basecamp, discusses the company’s approach to product strategy, including when to undertake a major product overhaul. Basecamp releases ground-up versions every few years when incremental updates reach a local maximum, and it supports older versions indefinitely to avoid forcing customers onto its schedule. Fried argues that a chat-first communication mindset can be toxic and that Basecamp’s tools are designed for asynchronous, organized communication rather than real-time chat. Despite generating tens of millions in revenue and profits, Basecamp deliberately keeps its team small to avoid the pressure of a large organization and to focus on creating great work. Fried advocates for careful growth over rapid scaling, a philosophy he plans to explore in a future book titled “Regrow.”


### [Facebook's Julie Zhuo on product design - The Intercom Blog](https://yomu.fyi/post/facebook-x27-s-julie-zhuo-on-product-design-the-intercom-blog.md)
- Company: [Intercom](https://yomu.fyi/company/intercom.md)
- Published: Oct 13, 2026
- Type: Explainer

Julie Zhuo, Facebook’s VP of Product Design, recounts her evolution from engineering intern to leading design for the core Facebook app, including News Feed and the Like button. She defines good design as an experience that is enjoyable, useful, easy, and accessible, noting that adding functionality without increasing complexity is a persistent challenge. There is no fixed ratio of designers to engineers; the ideal mix depends on the stage of achieving product-market fit. Effective design principles should be controversial, as they articulate why a company made a particular choice over alternatives. When competing products offer similar utility and ease, the ultimate differentiator is the emotional response the product evokes in the user.


### [Bob Moesta on Jobs-to-be-Done - The Intercom Blog](https://yomu.fyi/post/bob-moesta-on-jobs-to-be-done-the-intercom-blog.md)
- Company: [Intercom](https://yomu.fyi/company/intercom.md)
- Published: Oct 13, 2026
- Type: Explainer

Bob Moesta, a principal architect of the Jobs-to-be-Done framework, explains that JTBD interviews focus on uncovering the actions and progress a buyer is trying to make, not opinions or product features. He describes the key forces driving customer switching—push, pull, anxiety, and habit—and notes that innovation opportunities lie at the edge where people want to be better, not with habitual users. Moesta introduces the concepts of big hire (purchase) and little hire (actual use), emphasizing that value is delivered in the latter and that both must be designed for. He addresses zombie revenue, which can arise when customers move on after the job is done, don’t understand the product, or when the product serves a different job than intended. For startups new to JTBD, Moesta advises removing their product from the interview equation and starting with competitor customers to understand the context and outcome independently of the product.


[Newer posts](https://yomu.fyi/posts/page/4.md) · [Older posts](https://yomu.fyi/posts/page/6.md)