In a quasi-experiment in one intro course, sections whose worked examples labeled what each step was for did better on quizzes right after learning, with no significant difference on later exams. Live coding, and Parsons problems in place of writing code, showed no measured difference in learning. Public GitHub notebooks rerun in their saved order finished without error about a quarter of the time; no study has measured how code that does not run affects what readers learn.
Four system design questions that come up often (enterprise RAG, an LLM serving stack, a customer-support agent, online A/B testing with a data flywheel), three Python problems (dynamic batching, sampling, backoff retries plus a token bucket), and a STAR skeleton for behavioral rounds. Each design question has a speakable answer and a self-check list, and all code was actually run.
The twelve projects side by side: which functions each has you write, which lecture section it uses, and what example the spec page gives you to check against; which projects share a method (enumeration, search, Markov chains, learning from data) and which need something the lectures do not teach. Includes three worked examples and five self-check questions. The course gives no difficulty ratings or hour estimates, and neither does this post.
Week 1 is about getting a program to derive new facts from known ones: propositional symbols and five connectives turn knowledge into sentences; model checking enumerates all 2^n truth assignments to decide whether the knowledge base entails a query; inference rules and resolution work on the sentences directly, converting to CNF and deriving the empty clause by contradiction; first-order logic cuts the number of symbols with constants, predicates and quantifiers. With time links for every section, plus the approach, common mistakes and self-checks for the Knights and Minesweeper projects (reference implementations folded away).
Week 2 is about probability: conditional probability and Bayes' rule turn evidence into probabilities; a Bayesian network uses arrows to split a joint probability into one conditional probability table per node, inference by enumeration is exact but slow, and rejection sampling and likelihood weighting approximate it with samples; a Markov chain describes a state that changes over time with a transition model, and a hidden Markov model adds a sensor model to work back from observations to states you cannot see. With time links for every section, plus the approach and checks for the Heredity and PageRank projects.
Week 3 covers three ways to write an optimization problem. Local search keeps a single state: hill climbing moves to a better neighbor and gets stuck at local optima, and simulated annealing accepts a worse move with probability e^(ΔE/T). Linear programming writes the problem as a linear cost function plus linear constraints and hands it to an existing solver. Constraint satisfaction problems shrink domains with node consistency and arc consistency (AC-3), then find a solution with backtracking search guided by the MRV, degree and least-constraining-value heuristics. Includes per-section time links, plus the approach, pseudocode, pitfalls and checks for the Crossword project (the reference implementation is folded at the end).
Week 4 covers three forms of machine learning. Supervised learning fits a hypothesis function to labeled data: classify by the k nearest neighbors, or draw a decision boundary with a set of weights (perceptron learning rule, logistic regression, support vector machines), then use loss functions, regularization and cross-validation against overfitting. Reinforcement learning uses Q-learning to update the estimated value of each state and action from rewards, with ε-greedy keeping some probability of random exploration. Unsupervised learning clusters with k-means. With time links for every section, plus the approach and ways to check your work for the Shopping and Nim projects.
Week 5 is about neural networks: a unit multiplies its inputs by weights, adds a bias and applies an activation function; gradient descent adjusts the weights one small step at a time, and backpropagation passes the error back so hidden layers can be trained; for images, convolution and pooling come first and an ordinary network follows, which is a CNN; for sequences, an RNN feeds its output back into itself. With time links for every section, plus the approach, common mistakes and checks for the Traffic project.
Week 6 is about getting computers to handle natural language: parse sentences with hand-written context-free grammar rules, then switch to statistics, from n-grams and Markov-chain text generation to a Naive Bayes classifier for review sentiment; turn words into vectors with word2vec, process whole sequences with an RNN encoder–decoder, and finish with attention and the Transformer. Includes section-by-section time links, plus an approach, common mistakes and ways to check your work for the Parser and Attention projects.
The last post in the series: one table reviewing what each of the seven weeks taught and which projects it had; in 2023 the course replaced Week 6 with a new version that reaches attention and the Transformer, while the other six weeks remain 2020 recordings; the lectures do not teach how to use today's language models. The second half links to this site's overviews of nine follow-up courses, by where you want to go. Includes three review self-check questions.
Looplane turns a coding-agent task into inspectable boundaries: native side effects cross Looplane tools, permissions, and sandboxing, while external runtimes retain their own loops and tools before returning a patch for Looplane audit. This article maps the planned 20-part series.
Looplane's native lane is controlled by AgentRunner: prepare a workspace, request a model turn, execute tool calls, append observations, and enter verification only when the model stops calling tools. Step, wall-time, repetition, token, and cancellation guards can terminate the run independently of the model. Protocol translation belongs to the next article.
This article follows one Looplane tool call through its mechanical execution boundary: `SafePathPolicy` for paths and symlink escape, fixed argv with `shell=False`, a sanitized subprocess environment, read-version hashes plus atomic replace for writes, and process-group cleanup at timeout. Permission policy, OS containment, and tool programs are reserved for later articles.
HW0 checks CS181 prerequisites in four problems — y=Xw solvability, optimizing an objective, reasoning about randomness, and OLS in Python. The problem that slows you down most is the gap to patch before HW1.
HW1 uses an 800,000-year ice-core temperature dataset across four problems: kNN and kernel regression, a geometric proof of least squares, basis-function regression, and a probabilistic derivation of ridge and LASSO, ending with a coordinate-descent LASSO implementation.
Week 0 is about search: DFS and BFS differ only in whether the frontier is a stack or a queue; A* adds the steps already taken to a heuristic estimate and guarantees the shortest path when the heuristic never overestimates; Minimax has the two players take the maximum and the minimum in turn, and alpha-beta pruning cuts branches that cannot change the result. With time links for every section, plus the approach, common mistakes and ways to check your work for the Degrees and Tic-Tac-Toe projects; the reference implementations are folded at the end of each project.
CS50 AI covers the introductory material of artificial intelligence in seven weeks and twelve Python projects: search, logical inference, probability, optimization, machine learning, neural networks, and language. In the first four weeks a person writes the problem down precisely; in the last three the computer learns from data. Lectures, notes, projects, and autograding are free and open to people outside Harvard. This post is the entry to the series: the mechanism each week teaches, an example you can work through, a prerequisite check, and a reading path.
None of the five mature coding agents use Python — pi/opencode/claude-code run on TypeScript, codex rewrote TS into Rust, omp bolted ~80k lines of Rust native crates onto its hot path. looplane still chose Python; the costs are startup performance and packaging, compensated by lazy imports, uv, and Cloudflare Sandbox.
Five analysts fan out in parallel within one superstep, so latency is max rather than sum; backtesting and reflection stand before synthesis, restricting the LLM to explaining evidence that already exists.
AG2 continues AutoGen's ConversableAgent model: agents collaborate through messages, while GroupChatManager selects the next speaker by round robin, manual choice, randomness, or an LLM.
DSPy replaces handwritten prompt strings with task Signatures, execution Modules, and Optimizers that compile better instructions and examples against a dataset and metric.
Haystack turns indexing, retrieval, generation, and evaluation into replaceable Components connected by directed-multigraph Pipelines; it fits Python teams that want RAG flows to be tested, versioned, and deployed as code.
LangChain v1 provides a high-level agent loop through create_agent, runs it on LangGraph, and treats tools, structured output, and middleware as its extension boundaries.
Pydantic AI models an agent as Agent[Deps, Output]: dependencies, tool inputs, and final outputs are typed, and model results must pass Pydantic validation.
Django combines data models, migrations, authentication, admin, forms, and security defaults into one system; using it well still requires understanding QuerySets, middleware, async boundaries, and production settings.
Scrapy separates crawling into the Engine, Scheduler, Downloader, Spider, Item Pipeline, and middleware; it fits high-volume, rule-driven HTTP crawling where you need control over scheduling, throttling, retries, and storage.
Selenium drives real browsers through standardized WebDriver sessions, making it useful for cross-browser workflows, existing test assets, and remote Grid capacity; it can render JavaScript applications, but it does not guarantee bypassing CAPTCHAs or other anti-automation controls.
Apify is not a single crawler. It packages scraping programs as Actors, saves reusable configurations as Tasks, triggers them with Schedules, and delivers results through Datasets. It fits teams that do not want to operate queues, schedulers, and workers, but Actor fees, compute, proxies, storage, and transfer all draw from the same platform budget.
Browser Use combines browser state, model decisions, and actions such as click, type, and extract into a repeatable loop. The open-source package favors custom tools and execution control; Cloud manages browsers, profiles, proxies, and concurrent work.
CrewAI (GitHub 57.4k stars, MIT, PyPI 11.6M weekly downloads) defines agents by role, goal, and backstory, then groups them into crews for collaboration. Unlike LangGraph's graph-first and MAF's workflow-first approach, CrewAI is team-first — you don't draw nodes and edges, you describe who's on the team and what each person does. It fully removed its LangChain dependency in late 2024 and is now a standalone framework. The commercial side splits into the open-source package and AMP, a managed platform adding visual building, deployment, tracing, and compliance.
Firecrawl puts single-page scraping, site discovery, whole-site crawling, and JSON extraction behind one API. Cloud removes browser, proxy, and worker operations; self-hosting gives infrastructure control, but not the complete Cloud feature set.
LlamaIndex (51,775 GitHub stars, MIT, verified 2026-08-21) has moved its center of gravity from indexing to Workflows: the standalone llama-index-workflows package pulls 2.81M weekly PyPI downloads, more than the 1.97M of the llama-index umbrella package itself. This post covers the core abstractions, the trade-off against hand-rolling a pipeline, and a hands-on test of its defaults on Traditional Chinese text — at the same chunk_size=1024, English fits 4,645 characters and Traditional Chinese only 1,332. Plus one fact you need before choosing: the TypeScript port is archived and unmaintained.
Scrapling puts HTTP, Playwright browsers, CSS/XPath extraction, and a Spider API behind one Python interface. Adaptive selectors save element properties and relocate a target by similarity after a layout change, but the output still needs validation.
Temporal is a durable execution platform (Server 1.31.2, Python SDK temporalio 1.31.0, MIT, verified 2026-08). What separates it from BullMQ / Celery isn't scale but the guarantee: a queue guarantees a message gets consumed, Temporal guarantees a multi-call process runs to completion. The price is that Workflow code must be deterministic — and LLM calls are inherently non-deterministic. This post covers how to resolve that tension and when the constraint isn't worth it.
AI Engineer coding interviews aren't identical to SWE — beyond LeetCode medium, you'll face ML-flavored problems (implementing a tokenizer, writing a batch inference pipeline, handling sparse matrices). Strategy: practice LeetCode medium to 70% pass rate, then spend remaining time on numpy/pandas operations, data processing pipelines, and ML-related programming problems.
Digital-native PDFs already contain readable text — what's missing is structure, and heuristics can recover it. PyMuPDF, pdfplumber, pypdf, and Tika do this with zero GPU and zero inference cost. The biggest selection trap isn't accuracy; it's PyMuPDF's AGPL-3.0 license.
Rewriting tool descriptions from soft suggestions to hard rules (whitelist + consequence explanation) eliminated the LLM's incorrect tool selection; adding skip_signal=True fixed vector store double-indexing.
notebooklm-py reverse-engineers Google's batchexecute RPC protocol, letting you programmatically control NotebookLM via Python / CLI / AI Agent — including audio, video, slides, quiz generation and more.
Celery is Python's go-to distributed task queue, using Redis or RabbitMQ as a broker to offload long-running work to the background. DaoDao's AI service uses it to handle async tasks like LLM feedback generation.
FastAPI is a modern Python web framework built on type hints — it auto-generates OpenAPI docs, supports native async, and delivers performance close to Node.js. It's the top choice for AI/ML services and the most worthwhile framework to learn in the Python backend ecosystem.
The MCP tool was returning a description field that caused 1,033 job listings to exceed the token limit. The fix: exclude description by default and add pagination.