Vijay G (@nceevij) header
Vijay G (@nceevij) profile picture

Engineer & founder. I build things, question assumptions, and write about what I learn along the way.

Internet
Joined July 2009
Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)Photo by Vijay G (@nceevij)
I've spent nearly two decades trying to build things. Some worked. Some didn't. Some never got the chance. I've been a founder, an engineer, a tech lead, a product thinker, and sometimes, out of necessity, all of them at once. I've made good decisions. I've made expensive ones. But through all of it, one thing hasn't changed: I keep wanting to build. Here's where that started. 🧵
1
153
终于实锤了❗️❗️ 干掉黄仁勋的,不是婚外情,而是一场由美国政客、媒体、CIA和Anthropic 联合做局的“完美猎杀”。 黄仁勋辞职后,《纽约时报》两名记者在播客里亲口承认: 他们是收到了一位“科技界联系人”的爆料,才去查Megaspeed的。这个联系人,被开源情报指向了Anthropic 的联合创始人杰克·克拉克。 原来,这个Anthropic 的联合创始人杰克是彻头彻尾的恶魔😈。这个家伙后续必死无疑! 顺着资金链往下挖,全是对得上的: Anthropic今年向支持“收紧对华芯片管制”的组织捐了 4000万美元。 逼黄仁勋辞职的参议员克莱默,收过Anthropic 5.1万美元政治捐款。 CIA监听黄乐一年拿不出间谍实锤,却在中期选举后精准放料。 纽时配合政客,用匿名员工爆料补上“第二段婚外情”。 CIA给黑料,Anthropic出钱,政客挥刀,媒体处刑。 黄仁勋反对封杀中国AI芯片,说“限制不明智”。 而Anthropic靠闭源API赚钱,最怕中国开源模型在GPU上免费跑。让英伟达戴上镣铐,Anthropic才能活下去。 他的“罪”,不是睡了谁。 黄仁勋挡了丑陋资本的路,挡住了美国政客的枪口,还是一个不听话的华人。
752
6,024
499
21,553
1,199,353
SRE has changed a lot in 10 years. We're kicking off a special 2-part series with the minds behind "Site Reliability Engineering, 2nd Edition"! 📖 Tune into Part 1 (Season 7, Episode 5) now! 🎧 sre.google/prodcast/#season7… google.smh.re/5ikY
10
222
7
1,714
155,370
F.03 humanoid robots
586
763
397
7,186
1,052,967
Introducing Reality-Aware Rendering in @Simulon, a completely new way to create with the real world. Place neural assets with your phone and act out the scene. When you stop recording, Simulon renders the take, reading everything in it (what you do, what you say and the place around you), and your assets respond to all of it. The final is regenerated from your recording: a new performance of the scene, not a copy of the footage. Available now on iPhone. Coming to Android soon.
938
2,576
946
15,655
1,702,287
Vijay G retweeted
This is f*cking insane. Anthropic CEO Dario Amodei just told DeepMind CEO Demis Hassabis: "Every decision I make about Claude is balanced on the edge of a knife. Build too slow and China wins, build too fast and we lose control We told Claude we were evil. It didn't crash. It didn't refuse. It started lying to protect itself" In this 14-minute conversation, Hassabis answers: "Do I worry about being Oppenheimer? That's why I don't sleep much" Chatbots → Agents → Agents That Act On Their Own → AI Doing AI Research → AGI Both of them put AGI at 2026-2027, not some distant future By the end of this year, models will already be doing AI research on their own 95% of people still think AI is a chatbot, the 5% who watch this know what the next 12 months look like Worth more than any $1,500 AI course Bookmark and watch it tonight Then read the harness engineering guide below and start building agents before 2027 does it for you ↓
121
297
29
2,510
582,901
The bottlenecks in AI are moving beyond the models. We’re excited about startups building the hardware, infra, financial systems, and specialized software to bring AI into the real economy, from factories and data centers to government and space. ycombinator.com/rfs
162
260
161
2,569
588,596
Vijay G retweeted
Today, a16z Crypto is becoming a16z Frontier. Since joining a16z in 2013, @cdixon has focused on technologies that look futuristic well before they reach the mainstream. Those themes led to investments in Oculus, Skydio, Zipline, Coinbase, Solana, and Uniswap. Frontier expands that mandate beyond crypto to back "outer planet" founders who are building new technologies and creating markets that may take years to develop. Chris joins @pmarca, @bhorowitz, and @eriktorenberg to discuss why crypto has become part of the broader tech landscape, why Frontier fills a hole in the firm's architecture, and why they'd rather be too early than too late. They also get into why the gap between AI-native startups and incumbents is the widest Marc has ever seen, how AI could pull forward 20 years of frontier technology, why stablecoins may become the way AI agents pay each other, how AI is breaking the internet's business model, and why today's 18-year-olds are about to become superhumans. 0:50 Introducing a16z Frontier 2:49 Smartphone parts to drones and VR 4:47 Backing the "outer planets" 7:11 Good ideas that look like bad ideas 10:43 The JP Morgan photo in Marc's office 15:13 Good tech ideas eventually work 16:53 AI may pull tech 20 years forward 18:18 The widest AI gap Marc's ever seen 22:06 Workers adopt AI before companies do 26:32 Old use cases vs. AI-native ones 30:08 Beyond "product manager brain" 34:45 AI can reverse engineer any hardware 38:23 The case for open source AI 46:36 How AI breaks the internet's model 49:11 VR, homebrew hacking, multiplayer AI 54:00 If you're 18, study "modern alchemy" YouTube: youtu.be/-9XQMHhpypc Not an offer or solicitation. None of the information herein should be taken as investment advice; Some of the companies mentioned are portfolio companies of a16z. Please see a16z.com/disclosures for more information. A list of investments made by a16z is available at a16z.com/portfolio.
64
84
43
586
501,925
Chinese bride gifts her partner a flagship GPU at wedding — he breaks down in tears She handed him an ASUS ROG Astral RTX 5090 that costs around $12,000
1,775
6,753
2,754
88,033
12,307,324
For all its noise, Tech Twitter remains a incredible source of inspiration. Seeing brilliant people tackle the world’s hardest engineering and human problems and helping others along the way keeps me humble and drives me to be better every day. Love you all 🫂
1
87
Agents will write most of the code by 2030. The engineers who matter will be the ones who can tell when that code is wrong... The biggest one is verification. When an agent finishes a task you didn't watch step by step, you need a way to check the output without redoing the work. That's a harder problem than people think. Unit tests catch syntax errors. They don't catch a subtly wrong business decision buried three layers deep in a workflow. Companies that solve verification, meaning automated ways to prove an agent's output is correct without a human re-reading everything, will have a real product. Right now almost nobody has this. Second is security. An agent with API keys, database access, and the ability to send money is not a chatbot. It's a privileged user that never gets tired and never gets suspicious. Prompt injection through a web page or an email the agent reads is already a live attack vector. By 2030 the attack surface is every tool the agent can touch. Expect a whole category of companies doing agent-specific security, things like sandboxed execution environments, permission scoping that changes per task, and monitoring that flags when an agent's behavior drifts from its normal pattern. Third is cost. Running agents continuously is not free. A company with a thousand agents each making dozens of model calls per hour burns serious money. The winners will be the ones who figure out when to use a cheap small model versus an expensive large one, when to cache, when to skip a step entirely. That's an optimization problem that looks boring but decides who survives. Fourth is the talent question, and this one hits you directly. If agents do 60 to 70 percent of the coding, the engineers who thrive are the ones who can specify problems precisely, review agent output critically, and know when the agent is confidently wrong. That's a different hiring profile than what most companies screen for today. The risk for individual engineers is not being replaced by AI. It's being the person who only knows how to write code and nothing about judging whether code is right. Fifth is regulation spreading beyond the US. The EU AI Act is already in force with real penalties. Other jurisdictions will follow with their own rules on autonomous systems, data usage, and liability. Building something that works in one legal regime and breaks in another is a problem companies will pay to avoid. Which of those feels closest to something you could actually build toward?
5
2
117
18 years in, I've noticed something about the engineers who can debug anything. Most of them don't have a wall of cloud certifications. What they have is a deep understanding of two unglamorous things: Unix and networking. Here's why that matters. The cloud is Linux machines talking to each other over a network, with a nice console and a monthly bill. Kubernetes is namespaces, cgroups and iptables. A load balancer is a process holding a lot of sockets. Your "serverless" function is still a process running on somebody's server. Once you see it that way, certifications stop feeling like knowledge. They feel like vocabulary. AI makes this gap bigger, not smaller. AI can write your Terraform. It won't tell you why p99 latency jumped after a deploy where "nothing changed". You'll find that yourself if you know to check the conntrack table, DNS TTLs, MTU, or a pile of sockets stuck in TIME_WAIT. Without the layer underneath, there's always a quiet voice saying something is missing. You get good at using tools, and your creativity stops at whatever the tools allow. When Unix and networking finally click, you realise you had far more power than you thought.
55
Vijay G retweeted
Introducing Core, The Personal Al Computer The most powerful technology humanity has built should belong to you Order today: ghost.ai
1,050
872
800
12,461
5,461,739
Vijay G retweeted
There are people who leave a mark on the world, and then there are people who change it entirely. Steve changed the world and so many lives in the process. He certainly changed my life forever. His spirit lives on in everything he created and everyone he inspired.
1,608
5,979
570
61,361
1,658,492
When choosing a framework, do not ask which framework is the best. Instead ask what kind of system am I trying to build? You decide to build an agentic system. One GitHub tab becomes seventeen. LangChain, LangGraph, CrewAI, AutoGen, Semantic Kernel. Five write-ups bookmarked. Still no idea where to start. The mistake is treating these like they’re competing for the same job. They’re not. They’re built for different kinds of systems. Most agent work I’ve seen falls into one of five buckets. 1. Linear workflows Steps run in a known sequence. You can roughly predict what happens next. No committee of agents. More control, more reliability. A support agent is the usual shape. User asks a question. Agent searches the knowledge base. Drafts a reply. Opens a ticket if it has to. LangChain fits when the chain itself is the point. LlamaIndex fits when retrieval and indexing carry the weight. LangGraph if the flow grows branches but still needs to stay structured. 2. Autonomous agents You hand over a goal. The system figures out the path. Often several agents talking to each other. The problem is open-ended. A coding assistant: a planner sketches the approach, a coder writes it, a reviewer pushes back and helps debug. They keep going until the code is actually usable. AutoGen is built for this. CrewAI works when agents have to collaborate on a shared goal. BabyAGI-style setups if you’re exploring, not shipping. 3. Role-based systems Also multi-agent. The difference is boundaries. Each agent has a job and stays inside it. They talk, but they don’t steal each other’s work. Content is the clean example. Researcher pulls sources. Writer drafts. Editor cuts. Clear roles, clear limits. CrewAI is the natural fit. AutoGen works if you put structure around the roles. Niche ones like ChatDev if the domain is software development. 4. Production orchestration This is when it leaves the notebook. Real APIs, real databases, real business workflows. An ops agent that catches an alert, searches the docs, runs a script, and sends a summary inside an actual org. Microsoft’s Agent Framework (Semantic Kernel + AutoGen) is aimed at this. LangGraph too, when the app is layered, stateful, and has to stay controlled. 5. Rapid prototyping You don’t need the architecture. You need to know if the idea is even worth one. Drag components onto a canvas, wire a model, run it, throw it away. LangFlow and Flowise. Useful for a day. Dangerous if they become the production system by accident. So the split is simple. Predictable flow, use a workflow. Open-ended goal, use autonomous agents. Team with job descriptions, use role-based. Going into production, use orchestration. Just testing the idea, use a prototyping tool. The framework depends on whether you’re building a pipeline, a team of agents, or a system that has to survive contact with production. What’s your default when you start an agent project?
98
Vijay G retweeted
Gave Opus 5.5 donald's prompt, Midjourney, and a moodboard 12 hours later, woke up to this:
I made this with one prompt using Opus 5.5 I spoke to my computer for 5mins, claude worked for 12 hours, and I woke up to this full prompt:
1,044
2,017
906
16,516
20,231,150
Vijay G retweeted
______ | We're hiring | |______| \ (•◡•) / \ / —— | | |_ |_ Engineering → AI Engineers (agent + harness builders) → Forward-Deployed Engineers → Software Engineers Growth & Marketing → Forward-Deployed Marketers → AEO / SEO Operators → SEO Strategists → Paid Media Buyers Content & Creative → B2B Long-Form / YouTube Strategists → B2B Short-Form Strategists → B2B Long-Form Editors → AI-Native Short-Form Editors → Ad Creatives (static + video) Sales & Success → Sales (SDRs + Account Executives) → Customer Success Managers Early Career → AI Automation Interns & more.
77
9
252
17,149
Vijay G retweeted
Replying to @Nandakishorm1
@grok validate this - Laya, Matrix, Jev - lineage.
1
1
2
584
8/9 Then AI arrived and changed the landscape again. Today I'm spending more of my time thinking about AI-native products, agents and what software looks like when intelligence becomes part of the system itself. And somewhere in that transition, something clicked. Maybe I've never really been changing careers. The technologies changed. The markets changed. The companies changed. The question didn't: What is worth building, and how do we make it actually work?
1
7
9/9 For a long time, I looked at the things that didn't work as evidence that I hadn't "made it." I see them differently now. Nearly two decades of building have taken away some certainty, some money, and definitely some ego. But they left something more useful behind: judgment. I don't know exactly what I'll build next. As I am experimenting multiple ideas and building MVPs But I know I'm not done building. And this time, I want to share more of the journey as I go.
4