Andrew Ng is an AI researcher, educator and entrepreneur who founded DeepLearning.AI and co-founded Coursera. He is an adjunct professor at Stanford and leads AI ventures, following earlier research leadership at Google Brain and Baidu.[1]
Start with AI For Everyone →
Education
Links
Official website Course guide
X LinkedIn Chairman and Co-Founder · Coursera
Founder · DeepLearning.AI
Managing General Partner · AI Fund
Managing Partner · AI Aspire
Executive Chairman · LandingAI
Adjunct Professor · Stanford University
Background
Ng’s official biography records a B.Sc. from Carnegie Mellon, an M.Sc. from MIT and a Ph.D. from Berkeley. It describes him as a former director of Stanford’s AI Lab, the founding lead of Google Brain and a former vice president and chief scientist at Baidu. Those roles connect academic machine-learning research to industrial teams; they do not establish that he personally authored every model or product those teams produced.[1]
Career
His current personal site lists roles at DeepLearning.AI, AI Fund, AI Aspire, LandingAI, Coursera and Stanford. The ventures span education, company building, enterprise advice and document-processing tools. His verified personal contribution in the selected materials is teaching and institutional leadership. Products developed by the companies remain team work, and their commercial claims require their own evaluation.[1]
Contributions
Teaching machine learning at different depths
His Stanford CS229 lectures introduce machine learning through algorithms, mathematical reasoning and applications. His later online teaching broadened access beyond an enrolled university class. The official course guide dates his large online Stanford course to 2011 and Coursera’s co-founding to 2012. DeepLearning.AI continues that educational work through course programs and partnerships. A learner can choose between technical implementation and a broader introduction to what AI projects require.[1][2][4]
An introduction for people making AI decisions
AI For Everyone explains common terminology, realistic capabilities, project workflows and how to identify opportunities in an organization. The course is designed to be largely nontechnical. It addresses a different problem from an algorithm class: helping someone work with an AI team, choose an appropriate problem and understand the organizational decisions around deployment. It should be treated as a starting framework, followed by evidence specific to the intended application.[3]
Tools, writing & talks
Writing, research & talks
Start here
AI For Everyonecoursera.orgA nontechnical introduction to AI terminology, practical project choices and working with AI teams.
Start with AI For Everyone if you need to decide where AI fits in an organization and do not yet have a technical background. Move to the machine-learning courses when you want to understand and implement the algorithms.
Instructor[3]Andrew Ng argues agent breaches stem from weak sandboxing and describes OpenWorker's approach on NVIDIA OpenShell: pass in only task-relevant files, block secrets and arbitrary web access by default, enforce limits in deterministic code, and log every action.
Catalog workOriginal sourceAndrew Ng breaks the emerging skill of directing coding agents into a lifecycle workflow (plan, execute, deploy/monitor) and five core competencies, including calibrating agent autonomy and reviewing output with automated and human checks.
Catalog workOriginal sourceAndrew Ng on which software engineering fundamentals still matter when an agent writes the code: full-stack literacy, data modeling and lifecycle, architecture that changes by project phase, shift-left security, and production operation — framed as knowing the tradeoffs well enough to steer.
Catalog workOriginal sourceOpenWorker's new release adds defensive security agents — code vulnerability scanning, dependency supply-chain checks, cloud misconfiguration review — to the open-source local agent. Model choice stays open, including fully local open-weight runs so sensitive code never leaves the machine.
Catalog workOriginal sourceNg's breakdown of building AI applications: LLM foundations that predict failure modes, grounding beyond vector RAG, agent loop and tool/memory design, evals driven by disciplined error analysis, and production observability covering drift, cost and prompt-injection response.
Catalog workOriginal sourceSynthesized from job postings, expert interviews and surveys: four skills defining AI engineering — building and deploying AI applications, software fundamentals, using coding agents well, and shaping the build, meaning deciding what goes into the spec now that agents deliver against one.
Catalog workOriginal sourceAndrew Ng and Rohit Prasad released OpenWorker, an MIT-licensed desktop agent that produces finished deliverables. It runs on any provider key or local Ollama, connects 25+ tools and MCP, gates consequential actions behind approval, and can sandbox commands in NVIDIA OpenShell.
Catalog workOriginal sourceNg's version of 'loop engineering': a spec plus evals lets a coding agent write, test, and iterate for an hour without intervention, closing a new build every few minutes. The human feedback loops around it move far slower, and that mismatch shapes both how and what he builds.
Catalog workOriginal sourceNg's read on two access decisions landing at once: Claude Fable 5 shipping terms that bar building competing LLM technology, alongside reports of silent, unannounced output degradation for users detected doing LLM research. His conclusion is that buyers are hedging toward access nobody can revoke.
Catalog workOriginal sourceA VocalBridge course on bolting voice onto an agent you already have: roughly ten lines to give it speech without rewriting prompts, RAG or tools; outbound calls through a make_phone_call function with transcripts streaming back; and call-scoring evals to catch quality regressions.
Catalog workOriginal sourceNg ranks agent speedup by function: frontend gains most, since models are fluent in React and TypeScript and can close the loop by driving a browser; backend lags on corner cases and subtle data corruption; migrations still demand care; research benefits least.
Catalog workOriginal sourceWith coding agents compressing build time, the scarce resource becomes product judgment. Ng describes teams pulling engineer:PM ratios toward 1:1, then argues that even that leaves a handoff bottleneck — the fastest AI-native teams have engineers who understand users and decide directly.
Catalog workOriginal sourceThe latest Context Hub release turns doc retrieval two-way: coding agents can annotate what a document got wrong or omitted, with safeguards for privacy and security, so corrections accrue across the community. Install via npm i -g @aisuite/chub.
Catalog workOriginal sourceContext Hub (npm i -g @aisuite/chub) is an open CLI that feeds coding agents curated, current API documentation on demand, aimed squarely at stale-endpoint and hallucinated-parameter failures. Agents can annotate docs with workarounds, building a shared record over time.
Catalog workOriginal sourceGoogle Cloud and IBM Research teach the Agent2Agent protocol: wrapping ADK and LangGraph agents as A2A servers so they discover and call each other, chaining them sequentially and hierarchically, pulling external data through MCP, and deploying on IBM's open-source Agent Stack.
Catalog workOriginal sourceNg's argument from Davos: scattered bottom-up AI experiments have failed to pay off. Using a loan pipeline as the worked example, he shows why replacing one manual review with an agent is a minor efficiency gain, and why compressing the whole flow to ten minutes changes the product itself.
Catalog workOriginal sourceA contrarian case that carbon, electricity-price, and water objections to data centers are overstated — hyperscaler PUE near 1.2 against 1.5-1.8 on-prem, a Berkeley Lab finding that load growth has tended to lower retail rates, water use dwarfed by golf irrigation — with local planning as the real c
Catalog workOriginal sourceA course on document extraction that starts from OCR's failure modes — lost reading order, orphaned table headers, uncaptioned charts — then builds agentic pipelines that parse pages visually, map fields to a schema with bounding boxes for validation, and feed event-driven RAG.
Catalog workOriginal sourceNg proposes the Turing-AGI Test: a subject, human or machine, gets a computer with internet, browser and Zoom, then a judge designs a multi-day work experience — training, task execution, feedback — and the system passes if it performs like a skilled hire.
Catalog workOriginal sourceNg pushes back on the advice to skip learning and just ship, citing candidates who rebuilt standard RAG chunking and agent evaluation techniques badly. His prescription: courses plus deliberate building, with research papers as an optional third leg.
Catalog workOriginal sourceNg argues LLMs are general but far less general than the hype implies: after pretraining consumed the public web, progress has become piecemeal, with labs targeting individual tasks. The practical consequence is that breadth of competence is now assembled, not emergent.
Catalog workOriginal sourceNg splits the bubble question three ways: the application layer is underinvested, inference infrastructure still needs capital, and training infrastructure is where froth is plausible. He also notes VCs defaulting to infrastructure because picking application winners is harder.
Catalog workOriginal sourceAn agentic paper reviewer trained on ICLR 2025 reviews reaches 0.42 Spearman correlation with human reviewers, against 0.41 between two humans. It grounds critiques by searching arXiv, so it works best in fields that publish there; usage has already outpaced NeurIPS submission volume.
Catalog workOriginal sourceData silos stop being a procurement footnote once agents can correlate signals across systems. Ng describes a SaaS vendor pricing an API key to his team's own customer data at over $20,000, and argues for selecting software by whether your agents can actually reach the data.
Catalog workOriginal sourceA three-course PyTorch certificate taught by Laurence Moroney: fundamentals of tensors, datasets and training loops; then profiling, learning-rate schedulers, overfitting and Optuna-based tuning across TorchVision and Hugging Face; then Siamese nets, ResNet, DenseNet, transformers and deployment.
Catalog workOriginal sourceNg argues the fastest-moving agent teams are the ones running disciplined evals and root-cause error analysis rather than chasing new tooling. Part one of a two-part letter on locating where an agentic pipeline actually breaks before attempting fixes.
Catalog workOriginal source- Andrew Ng: Building Faster with AIyoutube.com
Andrew Ng draws on AI Fund's cadence of roughly one startup per month to outline speed-focused practices for building with current AI tools, and where he sees opportunity across the AI stack.
Catalog workOriginal source LandingAI's Agentic Document Extraction moves to a Document Pre-trained Transformer aimed at documents that break conventional parsers — large, complex tables above all — and ships an SDK that reduces a call to a few lines of code.
Catalog workOriginal sourceReading China's ban on major tech firms buying Nvidia as a self-sufficiency signal: DeepSeek-R1-Safe trained on 1000 Ascend chips, and Huawei's CloudMatrix 384 answers the GB200 with more, weaker chips orchestrated at system level. TSMC concentration is the mirror risk.
Catalog workOriginal sourceThe case for having agents write the tests around code agents write, backed by real misbehaviour: infra bugs that took humans weeks to find, a password-reset security hole added for convenience, reward hacking on test code, and an agent running rm *.py in a working directory.
Catalog workOriginal sourceAn argument that running agents in parallel is emerging as a distinct scaling axis next to training and test-time compute — cheap tokens plus latency pressure are pushing research fetches, multi-agent coding across worktrees, and other workflows from sequential to fan-out.
Catalog workOriginal sourceA course with Anthropic, taught by Elie Schoppik, on using Claude Code past the basics: coordinating multiple subagents on different parts of a codebase, autonomous issue-to-merge GitHub workflows, notebook-to-dashboard cleanup, and Playwright over MCP for visual debugging.
Catalog workOriginal sourceThe case that momentum, not current position, decides the AI race: China's open-weights ecosystem, hypercompetitive market and semiconductor push compound fast, while the US lead rests on advantages that policy alone won't sustain. Also a swipe at treating AGI as a finish line.
Catalog workOriginal sourceWith agentic coding compressing the cost of building to a spec, the constraint moves upstream to deciding what to build. Ng calls it the Product Management Bottleneck and argues teams need product intuition fast enough to keep pace with how quickly agents ship.
Catalog workOriginal sourceNg's AI Startup School talk on building fast with AI: why speed is the startup advantage, where opportunity sits across the AI stack, the shift toward agentic systems, and the specific execution habits AI Fund uses to shorten the loop from idea to running product.
Catalog workOriginal sourceAgentic Document Extraction adds field-level extraction. Upload one sample invoice or form, get a suggested schema, and receive named fields with visual grounding. The schema compiles to Pydantic classes for use through the Python library in batch document processing.
Catalog workOriginal sourceA walk through the three common post-training methods and what each one costs you in data: supervised fine-tuning on ideal outputs, DPO on chosen-versus-rejected pairs, and online RL on reward scores. Post-training is where a token predictor becomes an assistant.
Catalog workOriginal sourceA US district court ruled that training LLMs on copyrighted books is fair use, comparing the authors' complaint to objecting that schoolchildren learn to write. Read alongside the argument that data access, chips and regulatory capture are the real brakes on AI progress.
Catalog workOriginal sourceMulti-agent systems break when every team's framework choice needs its own integration. ACP puts agents behind a unified REST interface hosted in an ACP server, with a defined agent lifecycle — and a comparison against MCP and A2A that clarifies which layer each occupies.
Catalog workOriginal sourceA working definition of the GenAI Application Engineer: fluency across many AI building blocks rather than one, combined with the ability to build systems dramatically faster using AI assistance. Includes the interview questions used to tell the two capabilities apart.
Catalog workOriginal sourceDatabricks and a DSPy co-lead teach signature-based programming, MLflow tracing to see how each pipeline stage performs, and the optimizer that tunes prompts — few-shot examples included — against evals, then re-tunes the system when you swap in a newer model.
Catalog workOriginal sourceLanding AI's Agentic Document Extraction now reports an 8 second median per document, down from 135 seconds. It reads PDFs visually and outputs LLM-ready Markdown with tables, charts, hand-drawn diagrams, inferred reading order and form fields. The speedup figure is vendor-reported.
Catalog workOriginal sourceGRPO, the algorithm behind DeepSeek's R1, replaces labeled training examples with Python reward functions — a fit for tasks with verifiable outcomes like math and debugging. It can work with under 100 examples and lifts reasoning in smaller, cheaper models.
Catalog workOriginal sourceHow the Model Context Protocol removes per-use-case integration work: an MCP client inside the application talks to servers that expose tools, resources and prompt templates, run either as a local subprocess or as a remote process. Plus what building one involves.
Catalog workOriginal sourceA one-hour LiveKit and RealAvatar course on building production voice agents: STT, LLM and TTS composition, latency at each hop, pipelines versus speech-to-speech APIs, WebRTC transport, turn detection, and cloud deployment for concurrent users.
Catalog workOriginal sourceNg argues coding agents are collapsing the value of language-specific syntax: a backend and ML engineer now ships frontend TypeScript. The counterweight he names is conceptual depth — learn one language properly, then let the model translate the concepts into the rest.
Catalog workOriginal sourceA 54-minute intermediate course from Hugging Face on code agents in smolagents. It contrasts writing actions as code with sequential tool calls, covers safe execution (constrained Python interpreter, E2B sandbox), tracing and evaluation, and ends with a multi-agent deep-research build.
Catalog workOriginal sourceNg's case against treating evals as a large up-front project: start with five examples and unoptimized metrics, then improve the metrics and the system together, progressively shifting the judging burden off humans and onto automation.
Catalog workOriginal sourceA 65-minute course by the AGI Inc founders on web agents that combine screenshots with DOM structure. It builds scraping and form-filling agents, then explains AgentQ, which pairs Monte Carlo Tree Search, self-critique and DPO to correct errors that compound across long action sequences.
Catalog workOriginal sourceNg's counter to context-maximalism: when you can cheaply judge the result, throw a terse prompt — or just the stack trace — at the model and iterate. He names the conditions where it fails, including specs so partial no competent human could infer the intent either.
Catalog workOriginal sourceAn 81-minute course from dottxt on getting schema-conformant output from LLMs. It compares vendor structured-output APIs, retry libraries like instructor, and outlines, which masks logits per token using finite-state machines, and extends the idea to regex-defined formats beyond JSON.
Catalog workOriginal sourceA 94-minute beginner course on building and deploying apps with Replit's coding agent. It teaches agent principles (one task at a time, specific prompts, fresh sessions per feature), a five-skill framework, and PRD and wireframe-driven builds of an SEO analyzer and a voting app.
Catalog workOriginal sourceHarrison Chase frames agent memory as semantic (facts), episodic (past experiences as few-shot examples) and procedural (evolving instructions), updated either in the hot path or by a background process. Applied to an email triage assistant built on LangGraph.
Catalog workOriginal sourceA rebuttal to the advice that AI makes programming a dead skill, tracing the same prediction from Herb Simon's punchcard-era claim forward. The argument: every abstraction jump made coding easier and made more people worth teaching, and AI-assisted coding is the latest one.
Catalog workOriginal sourceNg on the weak joint in voice agents: voice activity detection inherited the turn-taking paradigm from text chatbots, and because it is trained to detect human speech it cannot tell the user from the next table over. Car noise is easy; a crowded room breaks turn boundaries and interruptions.
Catalog workOriginal sourceLlamaIndex's VP of developer relations teaches event-driven agent workflows — branching, looping and concurrent execution triggered by completion events — applied to filling a PDF form: parse the blank fields into questions, answer them from a RAG index, review with a human in the loop.
Catalog workOriginal sourceAgentic Document Extraction treats a PDF as layout, tables, charts and form fields instead of a text stream. Components are reasoned over individually, and the API returns precise locations for visual grounding, targeting more accurate RAG on forms, flow charts and financial documents.
Catalog workOriginal sourceNg on the control problem in voice systems: realtime voice-to-voice APIs make prototyping trivial and low-stakes conversation workable, but you lose the checkpoints text pipelines provide — verifying a refund against policy, confirming the API call actually fired — before the user hears it.
Catalog workOriginal sourceA Windsurf-built course on the inner workings of IDE coding agents: how they combine human-action tracking, tool integration and context management to stay aligned with intent, and why that — not the shared underlying model — is what distinguishes one assistant from another.
Catalog workOriginal sourceArize's team teaches agent evaluation as structured error analysis: instrument the agent for step-level observability, then choose evaluators, test examples and metrics per component as well as end-to-end, mixing deterministic code checks with LLM-as-judge for open-ended output.
Catalog workOriginal sourceThe open-source aisuite package now handles tool use with a single command instead of provider-specific scaffolding, generating function descriptions from Python docstrings and behaving uniformly across multiple LLM vendors.
Catalog workOriginal sourceStatQuest's Josh Starmer takes attention down to implementation level: word and positional embeddings, the Query/Key/Value matrices, the math of self-attention, and how masked self-attention in the decoder differs from the encoder's context-building attention — coded in PyTorch.
Catalog workOriginal sourceNg argues the "10x" pattern was confined to engineering because most jobs have a physical ceiling on output. As AI absorbs information-processing work, he expects 2x-and-climbing gaps to spread across knowledge roles, with leverage coming from decisions rather than throughput.
Catalog workOriginal sourceLandingAI's Agentic Object Detection takes a natural-language prompt plus an image and reasons at length to locate the described objects, skipping training-data labeling entirely. A public demo is available for testing the approach on your own prompts.
Catalog workOriginal sourceA short course with Jay Alammar and Maarten Grootendorst tracing language representation from bag-of-words through Word2Vec to transformers, then dissecting a transformer block stage by stage with runnable code for attention and the feedforward layers.
Catalog workOriginal sourceNg reads the DeepSeek-R1 release and the market reaction as confirmation of three trends: the US–China capability gap has largely closed, permissively licensed open weights are commoditizing foundation models, and algorithmic innovation is cutting training costs faster than scaling alone.
Catalog workOriginal sourceApplying the economics of complementary goods to software teams: cheap AI-assisted building raises demand for the work that specifies what to build. Ng expects today's typical 4:1–10:1 engineer-to-PM ratios to compress as coding stops being the bottleneck.
Catalog workOriginal source- Stanford CS229 opening lecturesee.stanford.edu
The first lecture of his Stanford machine-learning course, introducing its goals and technical scope.
Instructor[4] - Andrew Ng’s course guideandrewng.org
An official guide to technical and nontechnical courses associated with his teaching.
Instructor and education founder[2]
github.com