Long Long

After the release of iOS 26, I feel confident that AI is not replacing senior engineers any time soon. The progress of LLMs has been remarkable. Scaling compute and data has unlocked surprising capability. There are many well understood tasks that models can now handle, and that will continue to reshape the jobs market. But senior engineers spend most of their time on work that is not well understood. They operate in ambiguity. They make sense of new platforms, new APIs, new constraints. They combine broad experience with intuition to navigate the unknown. Most importantly, their context window grows over years when focused. Over the past few months I used ChatGPT 5 and Claude Sonnet 4.5 inside Xcode while updating Land of Livia to iOS 26. Both often sent me down dead ends with the confidence of a sociopath. Edge cases, platform inconsistencies, silent failures. The models could not reason about a new system based on limited training data. It’s not a matter of scaling data centers. That kind of reasoning requires approaches beyond LLMs. Until then, the frontier still belongs to humans.

42

7 Comments

New Frontier LLM Orchestrator Model Sakana AI! The fugu model series provides a novel paradigm for test-time scaling. The models are trained to call LLMs and infer not only which individual model is suitable for a given task, but also how to orchestrate queries across diverse multi-turn model query sequences. Thereby, it can provide more than just an argmax over a model pool. Throughout the last weeks, I have been extensively testing both fugu-mini and fugu-ultra, including day-to-day coding and within the ShinkaEvolve harness. All with great results! Check it out and sign up for the beta testing.

61

2 Comments

Shipped my first skill on skills.sh, a Guild.ai agent development skill for AI-native IDEs. 𝙒𝙝𝙖𝙩 𝙞𝙩 𝙙𝙤𝙚𝙨: any AI coding assistant (Cursor, Claude Code, Codex) can pull this in and start building Guild agents immediately. No need to read through docs manually or trial-and-error through SDK patterns. 𝘾𝙤𝙫𝙚𝙧𝙨: • SDK patterns (llmAgent, auto-managed, self-managed) • CLI workflow (init, test, save, publish) • Service integrations and tool selection • Runtime constraints and gotchas that aren't obvious 𝙒𝙝𝙮 𝙩𝙝𝙞𝙨 𝙢𝙖𝙩𝙩𝙚𝙧𝙨: building agents on Guild moves from "read the docs, write the code, hit the constraints, iterate" to "describe what you want, let your IDE scaffold it correctly the first time." That's the agent dev loop agentic engineers are starting to expect. Has a GitHub Action that checks weekly for doc changes so it stays in sync. 𝗜𝗻𝘀𝘁𝗮𝗹𝗹: npx skills add lalagias/guild-ai 𝗚𝗶𝘁𝗛𝘂𝗯: https://lnkd.in/d9S_C6_s  𝘀𝗸𝗶𝗹𝗹𝘀.𝘀𝗵: https://lnkd.in/dzdXZaYf Open to feedback if you try it out.

46

If you interview at Google for mid-to-senior roles, nearly every other round tests your system design skills. At Amazon? More than half of your interviews is about system design. At Microsoft? You can expect up to 60% of your rounds to go deep on architecture and design. At Meta? 70%+ of questions will challenge how you build and scale real systems. At Uber? Same story, system design is the backbone of the process. The more senior the role, the more system design dominates the interview, and with AI accelerating everything, even SDE-1s will soon need to master these fundamentals. Here are the most asked questions, so you can start preparing today, which will train you on the fundamentals that are continuously tested.  1. Social Networks & Messaging ↬ Design Twitter Feed ↬ Design Instagram Feed ↬ Design WhatsApp Messaging ↬ Design Facebook Messenger ↬ Design Twitter Direct Messages ↬ Design Instagram Stories ↬ Design Instagram Ads ↬ Design Twitter Notifications ↬ Design Tinder ↬ Design Reddit ↬ Design Facebook News Feed ↬ Design LinkedIn Connections ↬ Design Notification Service ↬ Design Facebook Messenger Payments  2. Media Streaming & Recommendations ↬ Design Spotify Playlist ↬ Design Netflix Playlist ↬ Design Netflix Recommendations ↬ Design YouTube Streaming ↬ Design Spotify Connect ↬ Design Spotify Discover Weekly ↬ Design Instagram Reels ↬ Design TikTok Feed ↬ Design YouTube ↬ Design Amazon Recommendations  3. E-commerce & Marketplaces ↬ Design Amazon Shopping Cart ↬ Design Amazon E-commerce Store ↬ Design Shopify Store ↬ Design Flight Booking System ↬ Design Stock Exchange System ↬ Design Payment System ↬ Design Uber Eats ↬ Design Uber Dispatch System ↬ Design Food Delivery App ↬ Design Uber Pool ↬ Design Uber Freight ↬ Design Booking System (BookMyShow)  4. File Storage, Collaboration & Productivity ↬ Design Google Drive ↬ Design Google Docs ↬ Design Dropbox File Sharing ↬ Design Distributed Cloud Storage (S3) ↬ Design Online Code Editor ↬ Design Code Deployment System ↬ Design File Sharing System ↬ Design Distributed Web Crawler  5. Infrastructure & Distributed Systems ↬ Design Distributed Message Queue (Kafka) ↬ Design Amazon SQS ↬ Design Distributed Locking Service ↬ Design Amazon EC2 ↬ Design Amazon DynamoDB ↬ Design Amazon Redshift ↬ Design AWS S3 ↬ Design Google Analytics ↬ Design Analytics Platform (Metrics & Logging) Check comments for more questions ↓ – P.S: You can find all of these problems on Layrs (layrs.me), It’s Leetcode but for learning System Design, we’re constantly adding more to our list. I quit my job at Google to build this platform and change the way we learn system design for every engineer out there. Also join our Discord Server, we’re quite active and would love to hear from you: https://lnkd.in/g9wGRadq

1,768

36 Comments

In our latest workshop, Alex J. Hong, from Syncly (YC W23), shared how brands can turn fragmented customer feedback into proactive, data-driven support. We covered: • Why visibility breaks as teams scale and signals scatter across tools • How Syncly’s CODA framework, Centralize, Organize, Discover, Act, creates a single source of truth • How AI transforms qualitative feedback into structured, actionable insights Thank you, Alex, for the insights! Full recap in the comments.

30

3 Comments

Felt cute, so I added 64-bit support for WASI preview1 in Wasmtime, the de facto standard WebAssembly runtime. https://lnkd.in/gtHxr34X Key points: 1. You can now compile a C program to wasm64-wasip1 using my fork of wasi-sdk (https://lnkd.in/gwqKX8NE) and run the resulting binary. 2. This feature is not standards-compliant so it will not be upstreamed. The WASI team is unlikely to agree to a memory64 preview1 ABI because they are focusing on preview2 and beyond. 3. With that said, this is useful if you want to run non-production workloads (e.g. for research purpose or to understand how memory64 can impact your application performance) High level steps: 1. Download the modified wasi-sdk toolchain (https://lnkd.in/g2YJt4Ys) 2. Compile your program targeting wasm64-wasip1 using the toolchain 3. Clone my fork of Wasmtime and checkout the yagehu/wasip1-memory64 branch 4. Build it with `cargo build --release` 5. Now you can run the wasm64 binary with `wasmtime run --wasm memory64 binary.wasm`

14

TLDR: Google Discover is getting an update today to make it easy for you to find, follow, and engage with the creators you care about. Read more at: https://lnkd.in/gzjXqsqk #google #ugc #x To see it in action⬇️

24

This post is a bit overdue (I'm writing it from my flight home from HIMSS 26), but it's one I've been wanting to share. On February 10-11, my colleagues Jin Lou, Jinyu Wang, Ashley Kim and I participated in an internal Agentic AI Hackathon at InterSystems. In just two days, our team delivered a fully functional proof-of-concept: an agentic AI app providing decision support for prior authorization case managers. It leverages RAG with vectorized payer policy documents so it can generalize and adapt to each payer's own sources of truth. After a week at HIMSS hearing from payers, providers, and vendors all converging on the same theme -- reduce administrative burden, redirect resources to care -- I'm realizing just how much of an opportunity this represents. Prior auth is one of the biggest friction points in healthcare today. Agentic AI isn't going to fix everything overnight, but proof-of-concepts like this show that the path from "interesting demo" to real-world impact is shorter than many people think. Huge thanks to InterSystems leadership for continuing to foster a culture where teams can explore ideas like this. More on my HIMSS 26 experience soon. #AgenticAI #HealthcareIT #PriorAuthorization #InterSystems #HIMSS26 #BurdenReduction

71

3 Comments

Google DeepMind produced their first AIGC short for the Sundance Film Festival. TLDR: a team of 45 animation veterans, researchers and engineers developed the look and produced “Dear Upstairs Neighbors” with DeepMind researchers using the artwork to train a new model to produce a symbiotic piece of media that requires a mix of art and science. Custom Training: Veo and Imagen models were trained on the team's original artwork and paintings. Creative control: The team created the story, then AI was used to transform rough animations into stylized videos. Precision editing: The technology allowed the artists to make specific edits without needing to recreate entire shots. You can read more at the link below (I was also impressed with the movie FLOW that was created by an even smaller team with the open-source 3D program Blender.) This tech are just tools to allow storytellers to produce what they see in their head faster and more efficiently. https://lnkd.in/gTgSzjTX

18

2 Comments

A tough week for thousands at Amazon. This week another round of layoffs was announced. While Amazon offers a generous package that gives people some time to figure out next steps, the reality is that for many the pressure is still very real. Families to support, visas that depend on sponsorship, and life plans that suddenly change. During the last wave in October, I read a lot of advice that felt genuinely helpful, and wanted to echo some of it again: - Reconnect with former teammates, managers, and people you've worked with before. Many opportunities start with a simple conversation. - Stay visible: share what you know and what you're looking for. - Write and share what you've learned (short posts, experience, lessons). Even small things help you stay present and keep learning. - Don't be afraid to reach out and ask for help if you need it. Asking for help isn't a weakness. - And just as important: take your time to process it. This is hard, and not everything needs to be solved immediately or in "productivity mode". If you're going through this right now: you're not alone. There are more people willing to listen and help than you might expect. And if you're on the other side today, with some stability: this is a good moment to be attentive, share opportunities, and lend a hand. If you think I can help in any way, feel free to reach out. Wishing strength to everyone affected.

320

Visuals speak louder than numbers. Inspired by the brilliant concept at clocks.brianmoore.com, we launched our own experimental benchmark to see how LLMs handle real-time coding tasks. The twist? We’ve expanded the lineup to include major Russian models. It’s fascinating to see how they stack up against global leaders in a pure visual comparison. No complex metrics or conclusions, just visualized outputs. Direct link: https://lnkd.in/gNPuSSTf Check out the original post below for more details! 👇

3

1 Comment

🚀🇹🇼 Welcoming the AWS Taipei Region! Big news for Taiwan’s tech and cloud community—Amazon Web Services (AWS) has officially launched the AWS Taipei Region! 🇹🇼 For over a decade, AWS has had a strong presence in Taiwan, but many of our clients had to rely on nearby regions like Singapore or Tokyo for their cloud deployments. Now, with local infrastructure, companies can enjoy: ✅ Lower latency for faster performance ✅ Data residency for compliance and security ✅ Stronger reliability with in-region redundancy This is a game-changer for Taiwanese enterprises, startups, and developers looking to scale with confidence. 恭喜 to the entire AWS team and the Taiwan tech ecosystem! Are you planning to migrate or deploy in the Taipei Region? #AWS #AWSTaipei #CloudComputing #TaiwanTech Source: https://lnkd.in/gYqvhhPH

2

2 Comments

Hot take: TUIs are having a renaissance because LLMs are primarily a text driven tools. Once LLMs break out of “the unreasonable effectiveness of text” we will see a swing towards other interface styles. AIs need to be humane interfaces. Computers needs to learn to interpret human communication, not force people to work within their limited communication methods.

5

2 Comments

🚀⚡️The Nova Act service on AWS is now live. Check out our no-code playground at nova.amazon.com/act to get started (no setup needed) 🚀 We’ve built the fastest way to go from prototype to production for UI automation workflows like QA testing, form filling, and data entry. Last week, we launched a SOTA model, new AWS service, and a frictionless developer experience. Our suite of tools supports the full developer journey: 1. Prototype in the no-code playground at nova.amazon.com/act 2. Build & iterate with the Nova Act IDE extension + SDK 3. Deploy, manage & operate your agents directly on AWS through the console A great builder experience is what will ultimately unlock adoption and scale for AI agents. Most agent builders require you to assemble primitives, wire tools, and manage infrastructure resulting in heavy lifting with slow payoff. With Nova Act, we took a different approach centered on accessibility and velocity, bringing everything you need into a single, integrated system. At the heart of the experience is the act API: act("book me a flight from BOS to SFO") Your agent moves through the browser, reasoning through each step — no scaffolding or specialized expertise required. We shaped Nova Act by obsessively iterating with customers every day. Those customers like Hertz, 1Password, Automation Anywhere, Katalon, First Orion, and Sola are now running the world’s first UI automation agents at scale in production. I’m incredibly proud of the team for bringing our agents vision to life from prototype to general availability in under a year. Their dedication to DevX for our customers from engineering to solutions shows in every detail of these tools! Can’t wait to see what you automate with Nova Act 🤗

18

2 Comments

Loved this post and these key points: 1. Engineers with product taste, and PMs with some technical experience 2. Less time planning, building/launching almost ready products and leaning into chaos

5

1 Comment

Excited to share a post I co-authored with Rong Hu on our large-scale migration to an in-house observability platform. We cover key lessons learned that we hope are useful for anyone tackling a similar challenge. https://lnkd.in/e6jfvWWv

78

3 Comments

Customer-centricity and product intuition are the foundation of building in the age of AI. The tools keep getting better, but the edge belongs to teams who deeply understand who they're building for. Running our own retail shop grounded our team in that — and built a real empathy for independent retailers that no dashboard or A/B test could.

8