Reference

Resources

71 sources in 8 sections. 64 were compared with the live page, 5 were not re-checked.

Sections

How to read the marks

Checked: The title and the content were compared with the live source on the date shown.
Not re-checked: Well known, but not re-checked. Verify it before you cite it.

Verification: items marked ✔ had their title and authors (papers) or page content (docs) checked against the live source on 2026-10-08 or, where an item says so, 2026-10-09. Items marked ◌ are well known but were not re-checked — verify before citing. Fast-moving docs (MCP, vLLM, provider APIs) change often; re-read the current version.

Foundations and architecture17

CheckedVaswani et al., Attention Is All You Need — https://arxiv.org/abs/1706.03762
Checked on 2026-10-09Wang et al., Self-Consistency Improves Chain of Thought Reasoning in Language Models — https://arxiv.org/abs/2203.11171
Checked on 2026-10-09Brown et al., Language Models are Few-Shot Learners — https://arxiv.org/abs/2005.14165
Checked on 2026-10-09Anthropic, Reasoning models don't always say what they think — https://www.anthropic.com/research/reasoning-models-dont-say-think
Checked on 2026-10-09Anthropic docs: structured outputs, extended thinking, effort — https://platform.claude.com/docs/en/build-with-claude/structured-outputs , /thinking , /effort
Checked on 2026-10-09OpenAI docs: structured outputs and reasoning guides — https://developers.openai.com/api/docs/guides/structured-outputs , /reasoning
CheckedWei et al., Chain-of-Thought Prompting Elicits Reasoning in Large Language Models — https://arxiv.org/abs/2201.11903
CheckedDao et al., FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness — https://arxiv.org/abs/2205.14135
Checked on 2026-10-09Sennrich, Haddow, Birch, Neural Machine Translation of Rare Words with Subword Units — https://arxiv.org/abs/1508.07909
Checked on 2026-10-09Holtzman et al., The Curious Case of Neural Text Degeneration (nucleus sampling) — https://arxiv.org/abs/1904.09751
Checked on 2026-10-09Petrov et al., Language Model Tokenizers Introduce Unfairness Between Languages — https://arxiv.org/abs/2305.15425
Checked on 2026-10-09Thinking Machines Lab, Defeating Nondeterminism in LLM Inference, 2025-09-10 — https://thinkingmachines.ai/blog/defeating-nondeterminism-in-llm-inference/
Checked on 2026-10-09NVIDIA, Mastering LLM Techniques: Inference Optimization, 2023-11-17 — https://developer.nvidia.com/blog/mastering-llm-techniques-inference-optimization/
Checked on 2026-10-09Meta, Introducing Meta Llama 3, 2024-04-18 — https://ai.meta.com/blog/meta-llama-3/
Checked on 2026-10-09tiktoken encodings (openai_public.py) — https://github.com/openai/tiktoken
Checked on 2026-10-09Anthropic, Claude Sonnet 5.5 migration guide (sampling parameters, tokenizer) — https://platform.claude.com/docs/en/models/sonnet-5-5/migration-guide
Not re-checkedJay Alammar, The Illustrated Transformer; Andrej Karpathy's "Neural Networks: Zero to Hero" videos

Retrieval and RAG10

CheckedLewis et al., Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks — https://arxiv.org/abs/2005.11401
CheckedMalkov & Yashunin, HNSW — https://arxiv.org/abs/1603.09320
CheckedGao et al., Precise Zero-Shot Dense Retrieval without Relevance Labels (HyDE) — https://arxiv.org/abs/2212.10496
CheckedLiu et al., Lost in the Middle: How Language Models Use Long Contexts — https://arxiv.org/abs/2307.03172
CheckedAsai et al., Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection — https://arxiv.org/abs/2310.11511
CheckedYan et al., Corrective Retrieval Augmented Generation — https://arxiv.org/abs/2401.15884
CheckedEdge et al., From Local to Global: A Graph RAG Approach to Query-Focused Summarization — https://arxiv.org/abs/2404.16130
Checked on 2026-10-09Cormack, Clarke, Büttcher, Reciprocal Rank Fusion outperforms Condorcet and individual Rank Learning Methods, SIGIR 2009 — https://plg.uwaterloo.ca/~gvcormac/cormacksigir09-rrf.pdf
Checked on 2026-10-09Anthropic, Introducing Contextual Retrieval — https://www.anthropic.com/news/contextual-retrieval
CheckedEs et al., Ragas: Automated Evaluation of Retrieval Augmented Generation — https://arxiv.org/abs/2309.15217

Agents, tools and MCP15

CheckedYao et al., ReAct: Synergizing Reasoning and Acting in Language Models — https://arxiv.org/abs/2210.03629
CheckedAnthropic, Building Effective AI Agents (prompt chaining, routing, parallelization, orchestrator-workers, evaluator-optimizer) — https://www.anthropic.com/engineering/building-effective-agents
CheckedModel Context Protocol, official docs and specification (current revision at check time: 2026-07-28; versioning page) — https://modelcontextprotocol.io/docs/learn/versioning.md and https://modelcontextprotocol.io/
Checked on 2026-10-09MCP 2026-07-28 specification pages: changelog, deprecated features, multi round-trip requests, transports, authorization, security best practices, tools, resources, prompts — base URL https://modelcontextprotocol.io/specification/2026-07-28/
Checked on 2026-10-09Anthropic, Tool use overview — https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview
Checked on 2026-10-09OpenAI, Function calling guide — https://developers.openai.com/api/docs/guides/function-calling
Checked on 2026-10-09Anthropic, How we built our multi-agent research system (2025-06-13) — https://www.anthropic.com/engineering/multi-agent-research-system
Checked on 2026-10-09Anthropic, Effective context engineering for AI agents (2025-09-29) — https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents
Checked on 2026-10-09LangChain, Context engineering for agents (2025-07-02) — https://www.langchain.com/blog/context-engineering-for-agents
Checked on 2026-10-09Chroma, Context Rot (2025-07-14) — https://www.trychroma.com/research/context-rot
Checked on 2026-10-09Yao et al., τ-bench — https://arxiv.org/abs/2406.12045
Checked on 2026-10-09Chen et al., Evaluating Large Language Models Trained on Code (pass@k) — https://arxiv.org/abs/2107.03374
Checked on 2026-10-09Sumers et al., Cognitive Architectures for Language Agents (CoALA) — https://arxiv.org/abs/2309.02427
Checked on 2026-10-09Park et al., Generative Agents — https://arxiv.org/abs/2304.03442
Checked on 2026-10-09Anthropic prompt caching docs — https://platform.claude.com/docs/en/build-with-claude/prompt-caching

Serving and efficiency9

CheckedKwon et al., Efficient Memory Management for Large Language Model Serving with PagedAttention (vLLM) — https://arxiv.org/abs/2309.06180
CheckedLeviathan et al., Fast Inference from Transformers via Speculative Decoding — https://arxiv.org/abs/2211.17192
CheckedvLLM documentation — https://docs.vllm.ai/
Checked on 2026-10-09Anthropic Message Batches API — https://platform.claude.com/docs/en/build-with-claude/batch-processing
Checked on 2026-10-09Anthropic, reducing latency — https://platform.claude.com/docs/en/test-and-evaluate/strengthen-guardrails/reduce-latency
Checked on 2026-10-09vLLM docs: optimization guide (chunked prefill, preemption), speculative decoding, automatic prefix caching — https://docs.vllm.ai/en/latest/configuration/optimization.html , /features/speculative_decoding/ , /design/prefix_caching.html
Checked on 2026-10-09Anyscale, Continuous batching (2023-06-22) — https://www.anyscale.com/blog/continuous-batching-llm-inference
Checked on 2026-10-09Zhong et al., DistServe — https://arxiv.org/abs/2401.09670
Not re-checkedOpenTelemetry GenAI semantic conventions (check current stability status) — https://github.com/open-telemetry/semantic-conventions-genai

Fine-tuning and alignment8

Checked on 2026-10-10Zhang et al., R-Tuning: Instructing Large Language Models to Say 'I Don't Know' (NAACL 2024) — https://arxiv.org/abs/2311.09677
Not re-checkedOuyang et al., Training language models to follow instructions with human feedback (InstructGPT) — https://arxiv.org/abs/2203.02155
Not re-checkedQi et al., Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To! — https://arxiv.org/abs/2310.03693
Checked on 2026-10-10Gekhman et al., Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations? (EMNLP 2024) — https://arxiv.org/abs/2405.05904
CheckedHu et al., LoRA: Low-Rank Adaptation of Large Language Models — https://arxiv.org/abs/2106.09685
CheckedDettmers et al., QLoRA: Efficient Finetuning of Quantized LLMs — https://arxiv.org/abs/2305.14314
CheckedRafailov et al., Direct Preference Optimization: Your Language Model is Secretly a Reward Model — https://arxiv.org/abs/2305.18290
CheckedHugging Face PEFT library docs — https://huggingface.co/docs/peft

Evaluation and safety10

CheckedZheng et al., Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena — https://arxiv.org/abs/2306.05685
CheckedOWASP Top 10 for LLM Applications, 2025 edition (Prompt Injection, Sensitive Information Disclosure, Supply Chain, Data and Model Poisoning, Improper Output Handling, Excessive Agency, System Prompt Leakage, Vector and Embedding Weaknesses, Misinformation, Unbounded Consumption) — https://genai.owasp.org/llm-top-10/
Checked on 2026-10-09OWASP LLM01 Prompt Injection — https://genai.owasp.org/llmrisk/llm01-prompt-injection/
Checked on 2026-10-09OWASP LLM05 Improper Output Handling — https://genai.owasp.org/llmrisk/llm052025-improper-output-handling/
Checked on 2026-10-09OWASP LLM06 Excessive Agency — https://genai.owasp.org/llmrisk/llm062025-excessive-agency/
Checked on 2026-10-10Stripe, idempotent requests (first response stored per key; keys can be pruned) — https://docs.stripe.com/api/idempotent_requests
Checked on 2026-10-09Anthropic, Mitigate jailbreaks and prompt injections — https://platform.claude.com/docs/en/test-and-evaluate/strengthen-guardrails/mitigate-jailbreaks
Checked on 2026-10-09OpenAI Moderation guide — https://developers.openai.com/api/docs/guides/moderation
Checked on 2026-10-09Simon Willison, The lethal trifecta for AI agents (2025-06-16) — https://simonwillison.net/2025/Jun/16/the-lethal-trifecta/
Not re-checkedTools to try: RAGAS, TruLens, DeepEval, LangSmith, Arize Phoenix, promptfoo (compare features on current docs)

Study aids on this site2

Top 100 priority questions · The core RAG pipeline · RAG interview drills and mock · Coding, project defense and mock rounds
Hands-on: Projects

How to use0

Read one paper or doc after each module's lessons. Write a 5-line note (problem, idea, result, limitation, interview angle). Rule: if you can't explain it without the page open, it isn't learned yet.