A decade in AI, distilled into the resources I actually keep coming back to — organized from beginner to advanced.

From training CNNs and RNNs, to the Transformer wave, to pretraining and fine-tuning, all the way to today’s RAG pipelines and AI agents — I’ve been through quite a few eras of this field. Along the way I’ve bookmarked more resources than I can count. Most of them? Honestly forgotten or half-read. But a handful have stuck — the ones I find myself coming back to on a random evening or a slow weekend, still learning something new each time. Here they are, organized by depth. Hope they’re useful to you too.

---

Beginner: No Background Required

These resources assume zero prior AI knowledge. They focus on building intuition and staying informed.

Social Media & YouTube

• 3Blue1Brown — Grant Sanderson’s visual math explanations are unparalleled. His neural network series is the single best way to build mathematical intuition for deep learning. Even after years in the field, I revisit these.
• Andrej Karpathy’s YouTube — His “Neural Networks: Zero to Hero” series builds a language model from scratch. Karpathy has a rare gift for making complex ideas accessible.
• Bilibili: 飞天闪客 — Excellent Chinese-language AI explainers with strong visuals. Great for Chinese-speaking beginners.

Newsletters

• The Batch (by Andrew Ng) — Weekly digest of the most important AI news, explained simply. Consistently high quality.
• Ben’s Bites — Daily AI newsletter. Quick, well-curated.
• The Daily Gradient — A curated bilingual (EN/中文) daily briefing on AI, Tech, Business and World affairs. Editorially synthesized from 35+ sources — great if you want one page that covers everything each morning.

Podcasts

• 硅谷101 — Chinese-language podcast covering Silicon Valley tech, frequently features AI topics.
• [a]16z AI podcast — Conversations with founders and researchers building AI products.

Self-Media / Content Creators

• 张Zara — Chinese AI content creator, explains trends and products clearly.

---

Intermediate: Some Domain Knowledge Needed

You’ve built a few models or at least understand the basics. These resources will deepen your understanding.

Newsletters

• Lenny’s Newsletter — Product management meets AI. Essential reading for understanding how AI products are actually built and shipped.
• Import AI (by Jack Clark, co-founder of Anthropic) — Deep, thoughtful weekly analysis of AI developments. One of the longest-running AI newsletters.
• Ahead of AI (by Sebastian Raschka) — Technical but accessible deep dives into ML research. Raschka is a gifted educator.

Podcasts

• 四十二章经 — Chinese tech and startup podcast with strong AI coverage.
• 张小郡商业访谈录 — In-depth Chinese-language business interviews, increasingly AI-focused.

Courses

• DeepLearning.ai Short Courses — Free, bite-sized courses on specific topics (RAG, agents, fine-tuning, prompt engineering). Perfect for busy practitioners who want to learn one thing at a time.

Platforms

• Papers with Code — Browse SOTA results by task, find implementations. The bridge between research papers and practical code.
• Hugging Face Blog — Model releases, technique explainers, and ecosystem updates.

Business & Industry

• Wall Street Journal — As AI reshapes the economy, WSJ has become an unexpectedly good source for understanding real-world AI impact — enterprise adoption, labor market shifts, regulatory developments. The business perspective matters.

---

Advanced: Deep Technical Background Required

These are for people who read papers, understand architectures, and want to push the frontier.

Stanford Courses

These are the gold standard. I keep revisiting them — 常看常新 (there’s always something new to learn):

• CS224N: NLP with Deep Learning — Chris Manning’s legendary NLP course. Comprehensive coverage from word vectors to modern LLMs.
• CS25: Transformers United — Guest lectures from researchers at the cutting edge. Each session is a masterclass.
• CS336: Language Modeling from Scratch — The newest addition. Build a language model from the ground up. Invaluable for understanding what’s really happening under the hood.

Full Courses & Specializations

• DeepLearning.ai Specializations — When you need comprehensive coverage of a topic (NLP, computer vision, MLOps), these multi-week courses go deep.
• Stanford CS229: Machine Learning — Andrew Ng’s classic. The mathematical foundations that everything else builds on.

Research Blogs

Reading directly from the labs:

• Anthropic Research — Safety research, mechanistic interpretability, scaling insights.
• Google DeepMind Blog — AlphaFold, Gemini, and foundational research.
• OpenAI Research — GPT series, DALL-E, alignment work.
• Transformer Circuits — Anthropic’s mechanistic interpretability thread. Fascinating deep dives into how neural networks actually work inside.

---

My Advice

Don’t try to consume everything. Pick one or two resources at each level that match where you are today. The goal isn’t to read everything — it’s to build a consistent learning habit.

And remember: the best resource is the one that makes you want to learn more. If a podcast gets you excited about AI while you’re commuting, that’s more valuable than a textbook collecting dust.

These are the resources that have stuck with me through a decade of rapid change. I hope they serve you well too.