Learn while you wait for a project

Learn the basics of AI and large language models through a task you already perform. Practice giving clear instructions, reviewing results and using AI responsibly. Complete the course and pass the assessment to earn an OpenAI Academy badge; this is not a professional certification.

Why we recommend it: You can brief an AI tool well and notice when its answer is wrong. On model evaluation projects you rate answers and flag errors.

About 70 minutes

Time

About 70 minutes

Needs

A ChatGPT account

Recognition

Digital badge after assessment

Learn how to use Claude for everyday work, organize tasks, work with projects and improve results. Complete the course to receive its completion certificate.

Why we recommend it: You can get a strong draft from Claude and tighten your prompt until it is right. On prompt and text writing projects you write prompts and note where the model falls short.

Completion certificate

Learn how to delegate to AI, describe tasks clearly, evaluate outputs and use AI responsibly. Take the concluding assessment to receive a completion certificate.

Why we recommend it: You can decide which parts of a job to hand to AI and check the result before you trust it. On model evaluation projects you score an answer against the guidelines and explain the score.

Completion certificate

Explore how AI works, where it is used and what it can and cannot do. A starting point that does not require a programming background.

Why we recommend it: You can explain in plain words what a model does and why it sometimes gets things wrong. On annotation projects you label data knowing what the model will learn from each label.

Learn the basic language of AI, practice writing prompts and try everyday tasks such as summarizing information and creating images.

Why we recommend it: You can follow the terms people use around AI and get a decent summary or picture out of a tool. On prompt and text writing projects you read the brief and write prompts that meet it.

An open access book on how speech is produced, represented and processed by computers, written as teaching material for engineering students.

Why we recommend it: You can read a spectrogram and tell vowels, consonants and silence apart. On speech and audio annotation projects you mark sounds, pauses and speakers accurately.

Learn how sound is analyzed and transformed, taught by Xavier Serra and Julius O. Smith III. The course has no fee and no paid certificate.

Why we recommend it: You can read the frequencies in a recording and name what you hear. On audio signal QA projects you find clipping, noise and dropouts.

5 weeks

Time

5 weeks, about 10 hours a week

Level

Intermediate

Learn how machine learning models work with audio and speech. The course expects a background in deep learning and transformers, and gives a certificate when you complete the exercises.

Why we recommend it: You can run a speech recognition model on your own clips and read where it fails. On speech and audio annotation projects you check recordings and labels against what the model needs.

Level

Background in deep learning

Recognition

Certificate for completed exercises

Build a voice agent that listens and answers in real time, in seven lessons made with LiveKit and RealAvatar. DeepLearning.AI offers it for free for a limited time.

Why we recommend it: You know where the delay comes from in a spoken conversation with a machine, and you can measure it. On model evaluation projects you judge how well a voice agent hears and responds.

About 1 hour

Time

About 1 hour

Level

Intermediate

Short courses on creating and editing video with generative AI. You enroll in each course for free on the Academy page.

Why we recommend it: You can get a usable clip from a text prompt and see what the model got wrong in it. On model evaluation projects you spot artifacts and rate how well a clip matches its prompt.

Step-by-step guides to AI avatars, voices, localization and video creation, open to read on the HeyGen site.

Why we recommend it: You can make a talking presenter clip and dub it into a second language. On deepfake and media integrity projects you check lip sync, voice and translation in a generated clip.

A community-driven course on computer vision, with a unit on video. It expects some Python and machine learning.

Why we recommend it: You can train a model to find objects in images and see why it misses some. On video annotation projects you draw and track objects the way a model needs them.

Level

Some Python and machine learning

Get familiar with IELTS through official practice materials, short courses and expert webinars. Start with the free resources and see where you need more practice.

Why we recommend it: You can read a long text under time pressure and write a clear answer in English. On text and NLP annotation projects you read long guidelines and follow reviewer feedback.

Learn to write clear, short and precise sentences. The pre-class lessons are public and work on their own, without the live class.

Why we recommend it: You learn to write clear, short and precise sentences. Label definitions, review comments and prompts all depend on that.

How search engines find and show pages, and what helps people find yours, from Google's own Search documentation.

Why we recommend it: You can write a page title that says what the page is and earns the click. On copywriting and SEO projects you write titles and descriptions that match real searches.

Write and run Python as you learn. Build your understanding of the language through interactive lessons and exercises.

Why we recommend it: You can automate a dull file task and debug it when it stops. On workflow automation projects you rename files and check their formats with small scripts.

A developer course from Anthropic Academy for people who want to build with Claude. The course is free. Its examples run on a Claude API key with a small amount of prepaid credit.

Why we recommend it: You can send a request to a model from your own code and read the reply. On prompt and agent design projects you test prompts from code and compare what comes back.

Short, practical lessons in Python, data visualization, Pandas and machine learning. Kaggle says the courses are at no cost and that you can earn certificates.

Why we recommend it: You can open an unfamiliar dataset and answer a question from it. On data science and analysis projects you clean a table, plot it and check a model's results.

A fast, practical introduction to machine learning with interactive exercises and programming tasks. The first module takes about 80 minutes.

Why we recommend it: You can explain overfitting and why a skewed training set gives bad predictions. On QA projects you spot labels and records that would mislead a model.

A free course for people with some coding experience: nine lessons of about 90 minutes each. fast.ai asks for about a year of coding and basic high-school math.

Why we recommend it: You can train an image classifier and work out why it gets examples wrong. On AI/ML engineering projects you judge which examples help a model and which ones hurt it.

Time

Nine lessons, about 90 minutes each

Needs

About a year of coding

Explore how organizations approach AI responsibly, including the principles and practical decisions involved in building trust.

Why we recommend it: You can spot bias in an AI system and say who it hurts. On model evaluation projects you flag unfair or unsafe answers and explain why.

A ten-week course for lawyers and law students on how software, data and algorithms work. The material is free through OpenCourseWare; assignments and certificates go through edX.

Why we recommend it: You learn how software, data and algorithms work, in a course made for lawyers and law students. Legal experts use that knowledge to judge what AI says about law.

Time

Ten weeks

Lecture videos and notes on clinical data, risk stratification, disease progression, diagnosis and clinical workflows, published under a Creative Commons license.

Why we recommend it: You can read a study of a medical model and see what its numbers leave out. On clinical data review projects you check what an AI says about a patient against the record.

Learn the foundations of cybersecurity and ways to protect people and organizations from common threats.

Why we recommend it: You can name the usual attacks and secure your own accounts and files. On security research projects you sort reports by the threat they describe.

Try guided security exercises in a practical learning environment. Choose from the rooms available without a paid subscription.

Why we recommend it: You can follow an attack step by step on a live machine. On security research projects you recognize a threat and label it correctly.

Learn cybersecurity basics to protect your own digital life and see the biggest security challenges organizations face. Self-paced and available in 20 languages.

Why we recommend it: You can spot a fake login page and explain how an attack spreads. On security research projects you tell phishing, malware and other threats apart.

Free online training in web security with interactive labs, from the creators of Burp Suite. An account is only needed to track your progress.

Why we recommend it: You can find a SQL injection or cross-site scripting flaw and show how it works. On security research projects you check whether an AI answer about a web attack is correct.

Complete Introduction to AI and earn a digital certificate that records your learning.

Why we recommend it: You finish Introduction to AI and get a certificate you can add to your Relay profile.

Learn the basics of process improvement, problem-solving and Lean Six Sigma. Complete the training and assessment to earn the White Belt certification.

Why we recommend it: You can map how a piece of work moves from hand to hand and see where it goes wrong. On QA projects you log errors and find the step that caused them.

Self-paced training in cloud computing and AI, with videos and hands-on practice. You register with an email address and no credit card, and you earn shareable digital badges.

Why we recommend it: You can find your way around cloud services and explain what each one is for. On DevOps and monitoring projects you watch cloud storage and the services around a model.

A learning plan of 10 hands-on trainings, about 10 hours in all, on evaluating AI models, building assistants and setting guardrails. Complete every assignment to earn a digital badge through Credly. Enrolling takes a free Skill Builder account.

Why we recommend it: You can set limits on an AI assistant and prove they hold. On model evaluation projects you test answers against a rule and record where they fail.

A four-hour intermediate course on the principles, techniques and best practices for designing effective prompts, and on guarding against prompt misuse. Enrolling takes a free Skill Builder account.

Why we recommend it: You can get a model to reason step by step and resist a user who tries to trick it. On prompt and agent design projects you write a prompt, test it and close the gaps it leaves open.

A one-hour introduction to agentic AI, focused on autonomous software agents that perceive, reason, act and learn within dynamic environments.

Why we recommend it: You can tell a chatbot from a system that takes actions by itself. On model evaluation projects you follow an agent's steps and mark where it went wrong.

A one-hour introduction to autonomous robots and robotic systems.

Why we recommend it: You can describe how a machine senses a room and decides where to move. On egocentric task demonstration projects you record the movements a robot has to learn from.

A 90-minute course on MobilityGen, a toolset built on NVIDIA Isaac Sim for generating high-fidelity motion data to train and evaluate mobile robots.

Why we recommend it: You can say what a robot needs from a sensor log before it can learn anything. On depth and IMU capture projects you check whether a recording is clean enough to train on.

A two-hour introductory lab on the fundamentals of working with Universal Scene Description (OpenUSD).

Why we recommend it: You can open a 3D file and read how its layers fit together. On 3D and LiDAR annotation projects you read a scene and find the objects you have to label.

A four-hour course that goes from single API calls to full agentic systems with grounded retrieval, deep planning and safe execution, using NVIDIA NemoClaw and NVIDIA OpenShell as the reference stack and secure agent runtime.

Why we recommend it: You can put an agent in a sandbox and control what it touches. On security research projects you check whether an agent stays inside the limits it was given.

A two-hour course on the fundamentals of building applications with the Omniverse Kit SDK.

Why we recommend it: You can put together a small app that opens and shows a 3D world. On rig and hardware operation projects you can follow how the team turns your capture into a simulated scene.

A two-hour course on creating an Omniverse extension that uses AI to generate realistic 3D materials from natural language descriptions.

Why we recommend it: You can turn a sentence like “wet asphalt” into a surface on a 3D object and judge whether it looks right. On LiDAR and point-cloud capture projects you notice the glass and wet surfaces a scan is likely to get wrong.

A two-hour lab on software-in-the-loop for testing and validating AI-driven robots in a digital twin environment.

Why we recommend it: You can run a robot’s software in a simulated room and read why a test failed. On multi-sensor rig capture projects you see why every sensor has to start on the same clock.

Record and edit audio on your own computer and see the waveform of what you recorded. Audacity says it has always been and will always remain free for everyone.

Try this: Record 30 seconds of speech in a quiet room, then again near a noise source. Compare the two waveforms and trim the silence from the better one.

An open source annotation tool for audio and video recordings, with several tiers of labels on one timeline. It runs on your own computer.

Try this: Open a short recording you own. Mark where each sentence starts and ends on one tier, and who is speaking on another.

Draw boxes and track objects across the frames of a video. CVAT has a free plan online, and its Community edition is free to run yourself.

Try this: Upload a short clip you filmed yourself. Draw a box around one object and track it across the frames, correcting the box where it drifts.

Edit, color and mix video on your own computer with the free version of DaVinci Resolve. Blackmagic Design also publishes training books and lesson files.

Try this: Import a short clip you filmed. Cut it to 15 seconds, add captions and check that each caption lines up with the speech.

A free translation memory application for Windows, macOS and Linux. It remembers what you have translated and can connect to machine translation services.

Try this: Translate two short paragraphs that share a sentence. See what the memory offers you the second time, and decide whether it still fits.

An open source platform for data labeling and AI evaluation, with a paid Enterprise version alongside it. The Community Edition is the one you can download.

Try this: Make a small project from ten short texts you wrote yourself. Label each one, label them again a day later and count where you disagree with yourself.

Practice English with an AI language tutor. The free Basic plan includes limited daily practice; more extensive access is available through a paid plan.

Why we recommend it: You can speak up in English without rehearsing every sentence first. On session moderation projects you guide a speaker through a recording in English.

A library of English lessons and practice that is free to use. A free account unlocks more; courses, IELTS preparation and tutoring are paid.

Why we recommend it: You can get through long instructions in English without stopping to look up every word. On annotation projects you follow a long English guideline label by label.

Learn Python fundamentals through practical exercises. Work with variables, lists, functions and NumPy as you begin exploring data. The first chapter is free; a paid plan opens the rest.

Why we recommend it: You can write a short script on your own. On data and pipeline engineering projects you read files, check counts and fix label exports.

Build the reasoning skills behind programming through interactive problems. Free access has daily limits; a paid plan offers more extensive practice.

Why we recommend it: You can break a fuzzy instruction into exact steps. On annotation projects you follow each step of a rule and catch the case it misses.

Refresh your understanding of financial statements and core accounting concepts. Selected lessons are free to preview; full course access is paid.

Why we recommend it: You can read a balance sheet and an income statement and see how they connect. On finance and economics projects you label line items and check figures in reports.

Explore text-to-speech and voice design. Practice listening for clarity, pronunciation and consistency across two short examples. The free plan has a limited monthly credit allowance.

Try this: Write two short instructions for a fictional customer-support task. Generate the speech and note which words are difficult to understand. Use material and voices you have permission to use.

See how text-based editing, transcription and speaker identification work. The free plan provides limited media processing and editing capacity.

Try this: Record a 30-second explanation in your own voice, transcribe it, correct three recognition errors and compare the final transcript with the original recording.

Create a short AI-presented video from a script. The Basic plan offers a limited free allowance and watermarked results.

Try this: Write a three-sentence explanation of a topic you know. Create a short presentation and review whether the speech, visuals and timing match your script.

Experiment with AI presenters and short videos using the free tier. Explore how a script, voice and visuals change the viewer's understanding.

Try this: Create a 20-second explanation for someone new to your field. Check whether the pace and presentation make the meaning easy to follow.

Turn a short script, article or presentation into a video. Pictory currently advertises 300 credits with its free trial and no credit card requirement. This is a trial, not a permanent free plan.

Try this: Write a five-line script about a practical AI use case. Convert it to a video, then identify one scene where the image does not support the script and revise it.

Explore AI-supported content workflows and how organizations measure their visibility in AI search. Trial access and available features depend on the current plan.

Try this: Draft a short explainer about a subject you know, verify any factual claims against reliable sources and improve one paragraph for clarity.

Transcribe and analyze interviews, meetings and recordings. Speak AI lists 100+ languages. Start with no credit card.

Try this: Upload a 30-second recording of your own voice, transcribe it, correct three errors and compare the transcript with the recording.

Audible offers introductory trials for eligible new members in participating countries. Audiobook access and whether you keep a title after cancellation depend on the plan and offer. See the terms shown in your country before joining.

Terms: Free trial for eligible new members. A payment method may be required. Membership renews at the stated price unless canceled; trial terms vary by country and plan.

A practical introduction to working with AI: how to use it as a collaborator, assess its answers and bring it into everyday tasks.

4 h 39 min

Understand how AI systems learn and where their strengths and limitations come from, through a lecture series you can listen to.

6 h 15 min

Explore how AI and other powerful technologies may change work, institutions and society.

12 h 7 min

Learn how AI changes the economics of prediction, decision-making and business strategy.

10 h 7 min

A reported account of OpenAI and the people, data, infrastructure and resources involved in building modern AI.

17 h 50 min

Practice using generative AI for everyday work, write more effective prompts and learn how to use the tools responsibly. Earn a Google certificate when you complete the course.

Why we recommend it: You can reword a weak request until the answer improves, and you know when to double-check it. On model evaluation projects you compare two answers and say which is better and why.

Explore workshops on prompting, AI tools, content creation and everyday workflows. Find the topic and format that match what you want to do with AI.

Why we recommend it: You can pick up an unfamiliar AI tool quickly and get something usable from it. On prompt and text writing projects you draft a prompt, test the output and improve the wording.

Work toward an IELTS test result that documents your English level. The test is booked and paid for separately from the free IELTS Ready preparation resources.

Why we recommend it: You can prove your English with a band score that employers and universities recognize. On translation projects you correct machine output and explain each fix.

Work with a tutor on the English you need for interviews, meetings and professional conversations. Choose lessons around your level and goals.

Why we recommend it: You can ask a clear question on a call and push back politely. On QA projects you discuss a reviewer's feedback and agree what to fix.

Build vocabulary, grammar and confidence through structured lessons, with business-focused content to support communication at work.

Why we recommend it: You can read a rule in English and know exactly what it asks of you. On transcription projects you follow the guidelines word by word.

Demonstrate your ability to prepare data, analyze it and communicate findings through exams and a practical assessment.

Why we recommend it: You show you can prepare, analyze and present data. On data projects you use it to check counts, spot outliers and report what changed.

Build the programming and data-handling foundations for machine learning, including Python and practical work with pandas.

Why we recommend it: You can load a dataset, remove the bad rows and get it ready for training. On data and pipeline engineering projects you check the labels before a batch moves on.

Develop financial modeling, valuation and analysis skills through practical coursework and assessments.

Why we recommend it: You can build a three-statement model from scratch and defend what a company is worth. On finance and economics projects you check a model's numbers, assumptions and reasoning.

Build skills in accounting, financial forecasting, analysis and reporting through a structured career track with assessments.

Why we recommend it: You can look at a set of accounts and say where the next year is heading. On finance and economics projects you check whether a forecast or report holds up.

Explore how to evaluate AI systems, identify opportunities in health care and plan implementation. Designed for leaders making decisions about AI in health care.

Why we recommend it: You can ask the right questions before a hospital trusts a medical AI tool. On physician review projects you judge whether an AI answer is safe and clinically sound.

Explore the legal, policy and governance questions surrounding AI, with a focus on understanding how organizations and institutions respond to the technology.

Why we recommend it: You can tell which laws cover an AI system and where they leave gaps. On legal practice projects you check a model's behavior against rules and guidelines.

Build practical financial models and work through valuation techniques using Excel-based examples and case studies.

Why we recommend it: You can set up a spreadsheet forecast and work out what a business is worth. On finance and economics projects you rebuild a calculation and find where an AI answer went wrong.

Study security foundations, risk and incident response, with practical work using tools including Python, Linux and SQL.

Why we recommend it: You can trace a breach through a system and write up what happened. On security research projects you review how an AI reads logs and rates risks.

Practice investigating alerts, analyzing evidence and documenting security incidents through a structured path for security operations.

Why we recommend it: You can work a shift on a security desk and tell a false alarm from an attack. On security research projects you sort alerts and mark which ones are real incidents.

Explore generative AI, Microsoft Copilot and ethical considerations through a structured learning path. Complete the path to earn its completion credential.

Why we recommend it: You can say what an AI assistant does well, what it does badly and what it should leave alone. On model evaluation projects you compare answers from tools like these and flag problems.

Study computer science and AI with Python through a professional certificate series that combines foundational programming with AI concepts and algorithms.

Why we recommend it: You can write real programs and reason about how fast they run. On AI/ML engineering projects you write a script, test it and fix what breaks.

Develop a structured approach to improving processes, finding root causes and carrying out improvement projects.

Why we recommend it: You learn to structure a process and find root causes. On QA projects you use exactly this to trace why a batch failed review.

An exam on the foundations of AI, machine learning and generative AI, for any role that works with them. It costs $100, takes 90 minutes and has 65 questions, at a Pearson VUE center or online.

Why we recommend it: You can explain how a large model gets tested and what responsible use looks like. On model evaluation projects you rate and compare model answers.

Eight beginner courses on using AI for brainstorming, research, writing, content creation, data analysis and building apps, with no AI or coding experience needed. Coursera lists $49 a month in the United States and Canada, with other prices elsewhere, and offers financial aid.

Why we recommend it: You can give a tool a clear brief and know when the result is good enough to use. On copywriting and SEO projects you edit a generated draft until it is ready to publish.

An exam on generative AI fundamentals, Google Cloud's offerings and techniques to improve model output, for anyone in any job role and with no prerequisites. Google's learning path prepares you for free; the exam costs $99 plus tax, takes 90 minutes and is valid for 3 years.

Why we recommend it: You can explain to a manager what a language model does well and why its answers vary. On model evaluation projects you see why one answer is stronger and say so in your review.