Carlos CaravanTsz @KamuiProtocol

Be like @poteto 😎 > Grew up in Singapore with no CS background. > Taught herself design as a teen. > Studied finance at Monash. > First program was an Excel VBA macro. > Failed a startup. > Started blogging and tweeting while learning to code. > EmberConf organizers found her tweets. > Invited her to speak in 2015. > She was a nobody in Melbourne. > Shaking through her first talk. > That one talk changed everything. > Landed at DockYard. Moved to the US. > Another talk got her into Netflix. > Built studio tools. Became a tech lead. > Then an engineering manager. > Missed building. Went back to IC work. > Open-sourced hiring-without-whiteboards. > Dan Abramov invited her to React at Meta. > Imposter syndrome. Said yes anyway. > Six years on the @reactjs team. > Sole author of the React Compiler Beta post. > Co-author of React Compiler 1.0. > Burned out. Took time off. > Started building agent tools for herself. > Joined @cursor_ai in 2026. > Taught agents to verify their own work. > That became pstack. > Cursor got acquired into @spacexai. > Rebuilt Grok Bot. > Now ships 2,000–2,500 PRs a month. > Went from 31K to 200K+ followers. > @elonmusk keeps reposting her. She didn’t start with a degree or confidence. She started by writing in public. Every major job came from people finding her posts. Now she’s building Grok Bot and still on the React Compiler core team. She is truly an inspiration. ✨✨

Go GOD LEVEL in SYSTEM DESIGN by learning these concepts NOW:

The Agent SDK is now available. Define agents as code, with prompts in Markdown and tools in TypeScript. Deploy them on our hosted infrastructure, or run them on your own. Built-in evals let you measure every change and make your agents better over time.

Prompt of the day: Split this "one-way door" PR into two: 1. A PR that implements everything, except it can be easily reverted (a two-way door) 2. A followup PR (stacked on top) that actually goes through the one-way door AKA expand/contract. Use feature flags as needed. --- Lets you de-risk a huge PR by testing large chunks of it in production first.

GITHUB JUST CREATED AN OFFICIAL CERTIFICATION FOR THE MOST IN-DEMAND DEVELOPER ROLE OF 2026. It is called Agentic AI Developer. GH-600. And it is the first formal signal that running AI agent teams is now a recognized engineering discipline with a credential behind it. Not a prompt engineer. Not a vibe coder. An Agentic AI Developer. The person who operates, supervises, and integrates AI agents across the entire software development lifecycle. The person who knows where agents fail in production. The person who understands how to build autonomous workflows that do not introduce catastrophic failure modes into CI/CD pipelines. The person every engineering team is going to need and almost none of them have right now. GitHub certifying this role changes the hiring conversation permanently. Before GH-600: "Do you work with AI agents?" is an interview question with no standard answer. After GH-600: the credential tells the hiring manager exactly what you know and what you can do before the interview starts. The engineers who get certified in the first wave of GH-600 will have a credential for a role that has more demand than supply for the next 3 to 5 years. The engineers who wait until it is mainstream will be competing with everyone who moved first. If you are already working with GitHub Copilot or building agent-driven workflows you are already doing this job. GH-600 is how you prove it. Bookmark this.

So uh...I guess it's time to announce tstack? I caved. I am publishing my skills now.

@theo @grok que se hizo?

@dbonneville Qué ocurrió? @grok

nerds, thank me later... systemdesign.io

AI Inference Engineer: The Roadmap │ ├── 01 Engineering Foundations │ ├── Python, C++ & CUDA Basics │ ├── Git, Linux & Profiling Tools │ └── Reading Framework & Kernel Codebases │ ├── 02 Deep Learning & Transformer Internals │ ├── Attention, KV Cache & Decoding │ ├── Tokens, Context Windows & Sampling │ └── Model Architectures: Dense, MoE & Multimodal │ ├── 03 GPU & Hardware Fundamentals │ ├── GPU Memory Hierarchy & Compute vs. Memory-Bound │ ├── CUDA, Triton & Kernel Optimization │ └── NVIDIA, AMD, TPU & Custom Accelerators │ ├── 04 Model Optimization │ ├── Quantization: FP8, INT8, INT4 │ ├── Pruning & Distillation │ ├── Speculative Decoding │ └── Compilers: TensorRT, torch.compile & ONNX │ ├── 05 Inference Engines & Serving │ ├── vLLM, SGLang & TensorRT-LLM │ ├── Continuous Batching & PagedAttention │ ├── Prefix Caching & Chunked Prefill │ └── Streaming APIs & Request Scheduling │ ├── 06 Distributed Inference │ ├── Tensor, Pipeline & Expert Parallelism │ ├── Disaggregated Prefill/Decode │ └── Multi-GPU & Multi-Node Serving │ ├── 07 Performance Engineering │ ├── Latency, Throughput & Cost Per Token │ ├── TTFT, TPOT & p99 Tail Latency │ ├── Benchmarking & Load Testing │ └── Profiling & Bottleneck Analysis │ ├── 08 Production Infrastructure │ ├── Docker, Kubernetes & GPU Orchestration │ ├── Autoscaling & Capacity Planning │ ├── Observability, Tracing & Failure Recovery │ └── CI/CD, Rollouts & Model Versioning │ ├── 09 Quality & Reliability │ ├── Accuracy Evals After Optimization │ ├── Regression Testing & Numerical Debugging │ └── Safety, Rate Limiting & Multi-Tenancy │ └── 10 Real-World Projects ├── Serve an LLM With vLLM and Benchmark It ├── Quantize a Model and Measure Quality vs. Speed ├── Write a Custom Triton Kernel ├── Build a Multi-GPU Serving Stack With Autoscaling └── Cut Cost Per Token by 50% on a Real Workload An AI Inference Engineer doesn’t just deploy models. They make models fast, cheap, and reliable at scale, squeezing every token out of every GPU. Profile. Optimize. Scale. Serve. One of the top job profiles right now.

The next great "language for agents" should be: - As fast as Rust - As little design space as Go - Have the strictest type/verification system ever written Bend seems like an interesting first effort

Una de las razones por las que ciertos programadores aún no se adaptan a la IA es que creían que su trabajo era escribir código. Pero lo más difícil de programar nunca fue la sintaxis. Fue entender sistemas que no se ven: cómo viaja un request, cómo se referencia la memoria, dónde vive el estado de una parte de la interfaz, cómo pasa un dato de un sistema a otro y cómo se divide y normaliza en el camino. Eso siempre fue el trabajo real de un programador. La sintaxis era solo el medio para llevarlo a cabo. Y esa parte, entender cómo funcionan las cosas, sigue igual de vigente. La diferencia es que ahora ya no lo escribes todo tú: tienes algo que lo escribe por ti. Así que si pensabas que programar era escribir mucho en el teclado, se entiende por qué la IA te molesta tanto.

Goose: my latest programming language. Safer, faster and less memory than Rust/C++ for algorithms where tightly packed, flat data structures matter. It can represent more kinds of data structures as flat, pointer-free bytes than any language I know. github.com/aardappel/goose 1/

@wvo Enviame un resumen y explicación de todo este thread (focus en Goose) @bot

. @kentcdodds on reusable workflows that keep working when you change AI agents.

@PawelJLisowski yeah i am constantly updating and simplifying pstack. the more capable models get, the less instructions they need. over time i expect the skills to shrink more and more, until they're basically just saved prompts

shh, week 1 sneak peek. don't tell @theo 🫣

There’ll be times where you over-guide the AI. It knows better than you and is more intelligent than you. There’ll be times where you under-guide the AI. It can go off-track, have the wrong biases, or miss context. How well you manage this superpower is the #0 skill I look for.

manager.shadcndesign.com/s/f_oG7dxmP528…

Una simulación de Nueva York generada en tiempo real. Puedes andar, ir en bici, conducir coche o helicóptero. Con efectos de luz, tiempo real, tráfico, reflejos y más... Todo en menos de 800KB y de código abierto. → mrdoob.github.io/slop-newyork/