Resources

A reading list I maintain for myself and anyone else who finds it useful. Heavy on AI, LLMs, crypto, security, and the places where software meets finance.

Last updated: February 2026

AI and LLMs

Papers worth reading

The ones I keep coming back to, roughly in the order you’d want to read them if starting from scratch.

• Attention Is All You Need — Vaswani et al., 2017. The Transformer paper. Everything since builds on this.
• BERT — Devlin et al., 2018. Bidirectional pre-training; set the template for fine-tuning that lasted years.
• Language Models are Few-Shot Learners — Brown et al., 2020. The GPT-3 paper. Showed that scale alone gets you surprisingly far without fine-tuning.
• Scaling Laws for Neural Language Models — Kaplan et al., 2020. Loss scales predictably with model size, data, and compute. Changed how labs plan training runs.
• Training Compute-Optimal Large Language Models — Hoffmann et al., 2022. The Chinchilla paper. Most models at the time were undertrained for their compute budget.
• Training language models to follow instructions with human feedback — Ouyang et al., 2022. The RLHF paper. A 1.3B InstructGPT beat the 175B GPT-3 on human preference.
• Constitutional AI: Harmlessness from AI Feedback — Bai et al., 2022. Anthropic’s approach to alignment using AI feedback guided by a written constitution.
• Chain-of-Thought Prompting Elicits Reasoning in Large Language Models — Wei et al., 2022. Asking the model to show its work actually makes it better at hard problems.
• Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks — Lewis et al., 2020. The original RAG paper. Combines what the model knows with what it can look up.
• GPT-4 Technical Report — OpenAI, 2023. Capabilities, safety evals, multimodal inputs.

Agent frameworks

• LangChain / LangGraph — The most widely used agent framework. LangGraph is the newer, graph-based approach for stateful agents.
• CrewAI — Multi-agent orchestration where agents have defined roles and collaborate. Independent of LangChain.
• Microsoft AutoGen — Conversation-based multi-agent framework from Microsoft Research.
• OpenAI Agents SDK — Released March 2025. Lightweight Python framework for tool use, handoffs, and guardrails.
• LlamaIndex — Best for document ingestion, RAG pipelines, and agentic workflows over your own data.
• Model Context Protocol (MCP) — Anthropic’s open standard for connecting LLMs to external tools and data. Adopted by OpenAI, Google, JetBrains, and others.

Memory and retrieval

• Letta (formerly MemGPT) — Stateful agents with self-editing persistent memory. The research project that became a production framework.
• Qdrant — Open-source vector database. Fast, good filtering, free tier.
• Pinecone — Managed vector database. Minimal ops if you don’t want to run your own.
• Weaviate — Open-source with built-in hybrid search (vector + BM25).
• Milvus — Built for billion-scale vector search.
• Chroma — Embedded-first. Good for local prototyping and small RAG pipelines.

Developer tools

• Claude Code — Terminal-native agentic coding. Understands full codebases, runs commands, manages git.
• Cursor — AI-first code editor forked from VS Code. 200K context window, codebase-aware completions.
• Windsurf — Agentic IDE with auto-context discovery. Good for monorepos.
• GitHub Copilot — AI pair programmer in VS Code, JetBrains, etc. Now includes autonomous agent mode.
• Hugging Face — Where the open-source models live. The Transformers library and LLM Course are good starting points.
• Weights & Biases — Experiment tracking and model registry. The standard for anyone fine-tuning or managing models in production.

Blogs and newsletters

• Simon Willison — Covers practical LLM usage and AI tooling with more depth and honesty than anyone else writing regularly.
• Lilian Weng — Long-form technical posts on architectures, agents, and RLHF. Her LLM Powered Autonomous Agents post is still a canonical reference.
• Chip Huyen — ML systems design, AI engineering, MLOps. Author of “AI Engineering” (O’Reilly, 2025).
• The Batch — Andrew Ng’s weekly AI newsletter. Good signal-to-noise ratio.
• Andrej Karpathy — His Zero to Hero course builds neural networks from scratch and ends with a GPT. Nobody explains this stuff better.
• Anthropic Research — Papers, model cards, and the Responsible Scaling Policy.

AI and Finance

Research

• From Deep Learning to LLMs: A Survey of AI in Quantitative Investment — March 2025. Comprehensive survey covering the full pipeline from predictive modeling to agent-based automation.
• FinGPT: Open-Source Financial Large Language Models — Lightweight, open alternative to BloombergGPT.
• FinRL: A Deep Reinforcement Learning Library for Automated Stock Trading — The paper behind the FinRL framework. DQN, DDPG, PPO, SAC across multiple markets.
• ACM ICAIF Proceedings — The main academic venue for AI in finance. The 2025 edition covers LLM market regime forecasting, adversarial RL for market making, and more.

Open-source trading tools

• FinRL — Deep RL framework for stock trading. Supports PPO, DDPG, SAC, and others.
• FinGPT — Open-source financial LLMs. Sentiment analysis, news summarization, return forecasting.
• QuantConnect / LEAN — Open-source algo trading engine. Python and C#, cloud or local.
• Backtrader — Event-driven backtesting in Python. Supports live trading through Interactive Brokers and Oanda.
• Freqtrade — Crypto trading bot with backtesting and ML strategy optimization.
• Zipline — The backtesting library that powered Quantopian. Still widely used for strategy research.
• QuantLib — The standard C++ library for derivatives pricing and risk management. Python bindings available.
• Awesome Quant — Curated list of quant finance libraries across Python, R, Julia, and more. Good bookmark.

Data APIs

• Alpha Vantage — Free tier (25 calls/day), extensive indicators, NASDAQ vendor. Good for research.
• Twelve Data — Real-time WebSocket ticks, multi-asset. Clean APIs.
• Polygon.io — Low-latency production data. Preferred for execution systems.
• yfinance — Python wrapper for Yahoo Finance. Free historical data. Research only, not production.
• Financial Modeling Prep — Fundamentals, DCF, earnings calendars. Generous free tier.
• FRED — Macro and economic time series from the St. Louis Fed.

Books

• Advances in Financial Machine Learning — Marcos Lopez de Prado. The practitioner textbook for ML in finance. Feature engineering, labeling, backtesting pitfalls.
• Machine Learning for Asset Managers — Lopez de Prado. Shorter companion covering clustering, feature importance, and portfolio construction.
• Python for Algorithmic Trading — Yves Hilpisch. End-to-end strategy development, backtesting, and deployment.
• Lopez de Prado’s lecture notes — Free slides from his Cornell course. Good alternative if you want the ideas without buying the books.

Blogs and communities

• Quantocracy — Aggregator of the best quant trading blog posts. Fastest way to see what the community is reading.
• QuantStart — Educational articles on algo trading, ML, and backtesting in Python.
• AI4Finance Foundation — The org behind FinRL and FinGPT. Worth watching on GitHub.
• Ernie Chan’s blog — Practitioner blog from the author of “Algorithmic Trading.” Recent posts cover AI and corrective ML for forex.

Blockchain and Crypto

Trimmed down from the original version of this page. Kept the stuff that still matters.

• Bitcoin Whitepaper — Satoshi Nakamoto’s 9 pages. Still the clearest explanation of how and why.
• Ethereum Whitepaper — Buterin’s 2014 vision for a programmable blockchain, with annotations on what changed since.
• Mastering Bitcoin — Antonopoulos. The technical deep-dive. 3rd edition (2023) is free under Creative Commons.
• Mastering Ethereum — Antonopoulos and Wood. Free, comprehensive, still the best single-volume reference.
• Ethereum Developer Docs — Official reference for EVM, gas, accounts, consensus, and smart contract standards.
• DeFi Specialization — Duke University — Campbell Harvey’s four-course Coursera specialization. The most rigorous DeFi education available.
• Uniswap Protocol Docs — Useful primary source for understanding how AMMs and DeFi liquidity actually work.
• The Complete Satoshi — All of Satoshi Nakamoto’s known writings, emails, and forum posts.
• Jameson Lopp’s Bitcoin Resources — Comprehensive and regularly updated.
• Trail of Bits Blog — One of the top smart contract audit firms. Covers blockchain security and exploit techniques.

Security

References

• OWASP Top Ten 2025 — The baseline web application security risk list.
• OWASP Top 10 for LLM Applications 2025 — AI-specific attack surfaces: prompt injection, training data poisoning, etc. Worth reading if you’re building anything with LLMs.
• PortSwigger Web Security Academy — Free, hands-on web security training with interactive labs. From the Burp Suite team.
• PortSwigger Research Blog — Where new vulnerability classes get published first. HTTP request smuggling, web cache poisoning, DOM clobbering all came from here.

Practice

• Hack The Box — The main competitive hacking platform. Active machines, CTFs, career-path labs.
• TryHackMe — More structured and beginner-friendly than HTB. Good for getting started.
• CTFtime — Directory for CTF competitions worldwide. Past challenges, upcoming events, team writeups.

Blogs

• Schneier on Security — Bruce Schneier on security, privacy, and policy. Been publishing for decades, still going.
• Krebs on Security — Brian Krebs. Investigative journalism on cybercrime and breaches.
• Daniel Miessler — Security, technology, and AI.

Cryptography

• Cryptography I — Dan Boneh — Stanford course on Coursera. Stream ciphers, block ciphers, MACs, public-key, digital signatures. Free to audit. The best intro available.
• A Graduate Course in Applied Cryptography — Boneh and Shoup. Free textbook. Graduate-level with formal proofs.
• New Directions in Cryptography — Diffie and Hellman, 1976. The paper that started public-key cryptography.
• NIST Post-Quantum Cryptography Standards — Finalized ML-KEM, ML-DSA, and SLH-DSA standards in August 2024. Required reading if you’re doing any crypto system design now.
• The Illustrated TLS 1.3 — Interactive byte-level walkthrough of a TLS 1.3 handshake. Best way to understand how transport security works in practice.

Cypherpunks

Kept these because the history matters.

• Cypherpunk Manifesto — Eric Hughes, 1993.
• Cyphernomicon — Timothy C. May.
• Cypherpunk Research — Tom Busby’s collection of primary sources.

Programming

• Go by Example — Annotated example programs in Go.
• Go Language Spec — The official specification.
• The Hitchhiker’s Guide to Python — Opinionated guide to Python best practices.
• Stack Overflow Blog — Still worth reading for the engineering culture pieces.
• Coding Horror — Jeff Atwood. Software engineering, culture, and the human side of building things.