Mark Hafen

[Matt Dancho (Business Science) @mdancho84]

Data Scientists: Your next career upgrade isn't another machine learning certification.

It's learning how to ENGINEER AI systems.

Right now, there are 2 types of people building with AI:

1. The API Caller
Knows how to connect to OpenAI, Claude, and other LLMs.

2. The AI Architect
Understands how LLMs actually work. And can build the systems behind them.

Both have value.

But the second skill set is a serious competitive advantage.

And Stanford just gave you a FREE way to learn it.

CS336: Language Modeling from Scratch (Spring 2026)

18 lectures covering the engineering behind modern LLMs:

→ Architecture: Tokenization, Transformers, MoE
→ Hardware: GPUs, TPUs, memory, FLOPs
→ Performance: Triton, FlashAttention, parallelism
→ Training: Scaling laws, data pipelines, distributed training
→ Deployment: Inference and evaluation
→ Post-training: SFT, RLHF, reinforcement learning

And here's the part I love.

You don't just WATCH someone build AI.

You actually:

✅ Build a Transformer from scratch
✅ Implement your own tokenizer and optimizer
✅ Write FlashAttention2 in Triton
✅ Engineer distributed training systems
✅ Turn raw web data into LLM training datasets
✅ Train a model to solve mathematical problems

This is the kind of hands-on engineering that separates knowing AI from understanding how it's built.

My advice to data scientists:

❌Stop collecting certifications.

✅Start building things that force you to understand the technology.

You already have a foundation in data, statistics, Python, and ML.

Now it's time to turn that foundation into AI engineering skills.

Stanford CS336. 18 lectures. FREE.

(YouTube playlist in the comments)

♻️ Repost for a data scientist who wants to become an AI engineer.

🚨 Want to learn how to build + ship AI and Data Science projects (that businesses actually want in 2026)?

On October 21st, I am hosting a free workshop to help you get started with AI + DS projects in Python (free).

Register here (500 seats): https://t.co/onpLpRwkzH

[Dipanshu Kushwaha @Dipanshu_AI]

Don't waste 2 years learning to become an AI agentic engineer in 2026.

Andrew Ng, the godfather of AI, gave the complete playbook to become one from scratch.

1 hour course. Free:

• 00:00 - AI agent basics
• 12:12 - AI Agentic workflows & design patterns
• 53:27 - Practical tips for building AI agents
• 1:20:30 - self-improving AI agent loops
• 1:30:19 - multi-agent AI systems

I watched it last night.

Halfway through, I realized I could get into Anthropic in weeks, not years.

Bookmark now. Watch it. Then build your own AI agent

[Max Kesselfieber @M_Kesselfieber]

The shame of what White women have done to their own civilization, their men, and their children will never be extinguished.

They are, collectively, cursed. They will be forever known as the witches who destroyed the West. Their future will be bleak...

[Dhanush N @Dhanush_Nehru]

AI Engineer: The Ultimate Learning Roadmap

├── 01 Python & Fundamentals
│ ├── Python, Git, Linux & APIs
│ ├── Data Structures & Algorithms
│ └── OOP & Software Engineering
│
├── 02 Mathematics for AI
│ ├── Linear Algebra & Calculus
│ ├── Probability & Statistics
│ └── Optimization & Gradients
│
├── 03 Data Engineering
│ ├── NumPy, Pandas & SQL
│ ├── Data Cleaning & Preprocessing
│ └── Feature Engineering & Pipelines
│
├── 04 Machine Learning
│ ├── Regression & Classification
│ ├── Clustering & Decision Trees
│ └── Model Evaluation
│
├── 05 Deep Learning
│ ├── Neural Networks & Backpropagation
│ ├── PyTorch or TensorFlow
│ └── Transformers & Attention
│
├── 06 LLM Fundamentals
│ ├── Tokens & Embeddings
│ ├── LLM Architecture & Context Windows
│ └── Pre-training vs Fine-tuning
│
├── 07 Prompt Engineering & APIs
│ ├── Prompt Design & Structured Outputs
│ ├── Function Calling & Tool Use
│ └── Cost Optimization
│
├── 08 RAG & Vector Databases
│ ├── Chunking & Embeddings
│ ├── Vector Search & Hybrid Retrieval
│ └── RAG Evaluation
│
├── 09 Fine-Tuning & Optimization
│ ├── SFT, LoRA & QLoRA
│ ├── Quantization & GPU Optimization
│ └── Fine-Tuning vs RAG
│
├── 10 AI Agents
│ ├── Tool Calling & Agentic Workflows
│ ├── Memory & State Management
│ └── MCP & Human Approval
│
├── 11 AI Security & Evaluation
│ ├── Hallucinations & Benchmarking
│ ├── Prompt Injection & Data Leakage
│ └── Guardrails & Testing
│
├── 12 MLOps & Deployment
│ ├── FastAPI, Docker & CI/CD
│ ├── Cloud Deployment
│ └── Monitoring & Model Versioning
│
└── 13 Real-World Projects
���── Build an AI Chatbot
├── Create a RAG Application
├── Develop an AI Agent
└── Deploy a Production-Ready AI System

AI engineering isn't just about calling an LLM API.

It's about building reliable AI systems that solve real problems in production.

Save this roadmap.
Learn one concept at a time.
Build projects as you go.

The goal isn't just to use AI. It's to engineer AI systems.

[Æsir @NS_Gottliche]

"I knew Hitler personally, I worked with him, and I am familiar with numerous details of his daily life, his methods, and his political and sociological conceptions. [...] Hitler did not believe in a personal god. He believed only in the bond of blood between succeeding generations and in a vague conception of fate or providence. Nor did he believe in a life after death. In this connection he often quoted a sentence from the Edda, that remarkable collection of ancient Icelandic literature, which to him represented the profoundest Nordic wisdom: “All things will pass away, nothing will remain but death and the glory of deeds.”

-- The labyrinth: memoirs of Walter Schellenberg, Hitler's chief of counterintelligence p.-93-94

[g2Ø @gh0st2offline]

it’s pathetic how we have time to scroll tt for 4h / day l, but not enough to read a 7min essay that could change ur life

anyway, here is a summary, before u decide to read the full thing

enjoy

[Zephyr @Zephyr_hg]

A masterclass on how to get FLUENT in Claude in one weekend, by building seven setups and then running them on your real work:

[Ge WEIMIN @GeWEIMIN8]

中俄通婚剧增，若不了解俄罗斯女性生理缺陷，婚后绝对会很痛苦！！！

从生理层面看，俄罗斯女性由于地域和遗传等因素，身体特征与国内女性存在差异。我见过太多这样的例子，男方一开始觉得特别有面子，可婚后没多久就开始抱怨。

我家远房��弟去年刚娶了个俄罗斯姑娘，办婚礼的时候整个家族都去凑热闹，堂弟带着高挑明艳的洋媳妇站在台上，一米八五的堂弟站人家姑娘旁边都没显得格外魁梧，当时多少去吃席的男亲戚，眼睛都直了，背地里羡慕堂弟走了大运，说小伙子这辈子都值了。结果才过了不到一年，堂弟上次偷偷找我喝酒，三杯白酒下肚就红了眼，说这日子过得实在难受，早知道说啥都不结这个婚。

堂弟说的第一个问题，就是体味问题。我之前其实也听过说欧美人体味重，可没想到会严重到这个程度。俄罗斯纬度高，天���冷，当地人自古就习惯吃高热量的红肉、奶制品，还爱喝高度酒，本身汗腺就发达，天生体味就比我们亚洲女性重很多。堂弟说，刚谈恋爱的时候都是在外面约会，姑娘天天喷好多浓香水，俩人也很少近距离待很久，他只觉得香水味有点浓，没往别的地方想。结婚住在一起才发现，香水盖不住底层的体味，尤其是运动完或者天气热一点的时候，那个味道混着香水，直往鼻子里钻，堂弟天生鼻子敏感，闻到那个味道就犯头疼，连着好几个月都睡不好觉，天天失眠。姑娘也挺委屈，已经做到一天洗两次澡，出门进门都补喷香水，遗传带来的东西，她也改不了��，为这事儿俩人吵了好多次，好好的感情磨得没剩多少了。

第二个问题就是皮肤衰老速度差太多。俄罗斯姑娘年轻的时候真的绝，皮肤白皙，轮廓深邃，十几二十岁的时候嫩得能掐出水，可她们松弛衰老的速度真的比国内女性快太多。堂弟媳妇今年才二十六，按说正是女人最好的年纪，可这两年嫁到国内，天天跟着我们吃重油重盐的饭菜，本身又爱喝甜奶茶，不到两年胖了快二十斤，胖了之后皮肤松得特别快，眼角的皱纹都出来了，颈纹也深得能放进去一根手指，跟结婚时候拍婚纱照比，简直像换了一个人。堂弟说，每次带她出去，朋友都背地里开玩笑，说你这哪里是娶媳妇，分明是带了个大姐出来，说者无心听者有意，堂弟心里别���多别扭了。

第三个问题说出来可能好多人想不到，就是生理周期和产后恢复的问题。我之前查过相关的数据，俄罗斯女性因为从小饮食结构的问题，普遍骨量更大，骨架也更宽，按说生孩子应该更顺利？其实根本不是，俄罗斯女性的生理期普遍比亚洲女性反应大得多，我堂弟媳妇每次来生理期，都疼得下不来床，要吃强效止痛药才能缓过来，国内的中成药调理对她根本没用，每个月都要有两三天躺平，家里所有活儿都扔给堂弟，时间久了堂弟也吃不消。

更麻烦的是产后恢复，去年秋天姑娘怀了孕，快到预产期去做检查，医生说孩子的个头比同孕周的亚洲胎儿大一圈，本来俄罗斯姑娘骨架大，我们都以为能顺，结果生的时候还是顺转剖，挨了一刀。产后更头疼，人家俄罗斯女性的体质就是生完孩子很快就能下地蹦跶，可我们这边的坐月子习俗她根本接受不了，忍了半个月就忍不住洗澡吹空调，结果落下了腰疼的毛病，现在一变天就疼得直哭。而且产后身材走样得特别厉害，肚皮松得收不回去，她自己也焦虑，天天跟堂弟吵架，说堂弟嫌弃她了。

其实好多人只看到娶俄罗斯媳妇新鲜有面子，觉得人家姑娘大方开朗不拜金，不要房不要车，可这些藏在表面底下的差异，根本没人提前告诉你。文化差异生活习惯差异都能慢慢磨，可这种生理上天生带来的差异，真的是怎么磨都磨不平，难受的只有自己知道。我不是说中俄通婚不好，也不是歧视俄罗斯姑娘，只是真的想提醒那些一心想娶俄罗斯姑娘的小伙子，别光盯着面子和新鲜感，一定要提前了解清楚这些问题，扪心问问自己能不能接受，别等结了婚才后悔，耽误了自己也耽误了人家姑娘。你们说，这种天生的生理差异，真的能靠感情完全克服吗？

[LinaHua @Linahuaa]

White people are the EASIEST prey for manipulation- and that's why their democracies are so dysfunctional, and that's why thet get dominated by Jews.

And it all has to do with
1) Christian roots and being friendly / trusting towards outsiders

2) Being a high trust society for longer than anyone else

3) Having less reason for manipulation, which means they're also less exposed to manipulation, which consequently means they lack experience to defend themselves against manipulation

In the arena of manipulation Asians and Jews are basically trained-from-childhood heavyweight MMA fighters.

While Whites are untrained fat neckbeards that get oneshot by one simple lie.

[AI Notkilleveryoneism Memes ⏸️ @AISafetyMemes]

The Intelligence Curse is happening

TLDR: When AIs make humans economically useless, governments and corporations have no incentive to invest in their people

For centuries, normal people could withhold their labor to negotiate. That era is almost over.

This breaks democracy - politicians won't need the support of the people - and leads to severe concentration of power

[Zherka @ZherkaOfficial]

Just fuck the stupid slut

[Ezennaya🧊 @Chrisslimz]

One of the painful parts of being a 30 year old woman is watching men who are same age with you date girls who are 18-23..

[Dragos Roua @dragosroua]

I lived in Asia (Korea + Vietnam) for the last 3 years and I'll try an explanation to this: basically, it's a cultural thing. Asia is a very pragmatic culture, and they are always looking for the shortest path. So, to put it very, VERY, bluntly: if it's working, it's beautiful. If it's not, it's ugly.

So your ideas of design and beautiful simply don't apply in that system.

Again, I'm oversimplifying, but this is just one of the many basic, cultural differences between Asia and the "West".

[Math Files @Math_files]

Fields Medalist Timothy Gowers raises a troubling possibility: AI could become so powerful at mathematics that future generations may lose the motivation to study the subject deeply enough to understand its discoveries.

The paradox is striking. Machines might eventually solve problems far beyond human capabilities, while fewer mathematicians possess the expertise needed to interpret their results.

For Gowers, the deeper concern extends beyond the future of mathematical careers. It is about preserving the curiosity, intellectual discipline, and satisfaction of discovery that have sustained mathematics for centuries.

He also envisions a possible transformation of the mathematician’s role, from discovering new theorems to interpreting, connecting, and explaining mathematical knowledge generated by AI.

Reflecting on the approaching social consequences, Gowers offers a vivid analogy:

“We’re still on the beach, we can see the tsunami, but it hasn’t quite reached us yet.”

Importantly, he is not calling for mathematics departments to close or for universities to stop admitting PhD students. His remarks are cautious predictions about a possible future, rather than descriptions of changes that have already taken place.

[Handre @Handre]

Rome collapsed because Romans stopped enforcing the conditions that made Roman citizenship worth anything. The same is happening to France right now.

The Western Empire's third- and fourth-century emperors, desperate for tax revenue and military manpower, settled entire Gothic and Frankish tribal confederacies inside the limes without demanding assimilation, military oaths, or any meaningful integration into Roman economic life. In 382 AD, Theodosius I settled roughly 100,000 Visigoths in the Balkans under their own chieftains, exempt from Roman law. They owed Rome nothing, and they acted accordingly. By 410, Rome fell.

The Austro-Hungarian Empire ran a version of the same experiment. Vienna absorbed millions of Slavic, Magyar, and Romanian subjects into a single fiscal and administrative unit without establishing coherent property rights or common rule of law across ethnicities. Capital fled the peripheral regions because contract enforcement varied by ethnicity and political connection. The empire spent itself policing internal tensions rather than producing anything.

Open borders can work when you couple them with private property, enforceable contracts, and zero access to a welfare state. Every historical collapse involved governments absorbing populations onto public land, public payroll, or public grain dole. Welfare states and open borders function as a single detonator, not two separate policies.

[Kevin Bass @kevinnbass]

They all do this. Meloni did it. Sanae Takaichi is flooding Japan with immigrants now too. Now Le Pen.

[Matt Van Horn @mvanhorn]

I don't think people realize the movement happening with "Reverse Engineer Anything," which just crossed 300k GitHub stars. It's not about "oh, let's make an open source Photoshop" or "the death of software." It's about unlocking human creativity. It's about what this UNBLOCKS in our brains, what new building blocks we have, and the big question: WHAT WILL YOU BUILD?! cc @reatools @morluto