Everything you can study here

The engineer who integrates artificial intelligence models, vector search, and LLM APIs into production applications. This course sequences Python scripting, backend APIs, and modern AI tool integration.

The practitioner who designs, evaluates, and optimizes structured prompts and workflow pipelines for large language models. This course sequences prompt engineering, Python API scripting, and LLM fundamentals.

The person who writes the half nobody sees: the part that holds the data, decides who is allowed near it, and keeps answering when a thousand people ask at once. This course assumes no programming, and it is ordered language, then runtime, then data, because a server framework learned before the language under it is a set of incantations, and a database learned before either is a filing cabinet with no reason to exist. Each unit ends in something you build, and the course ends in a service someone else could call.

The engineer who provisions, manages, and scales cloud infrastructure services. This course sequences cloud fundamentals, architecture patterns, and infrastructure security.

The specialist who secures computer systems, network perimeters, and cloud environments against threats. This course sequences scripting automation, security protocols, and cloud architecture.

The specialist who transforms raw database records into clear metrics, trends, and business intelligence. This course sequences database querying, analytical scripting, and data presentation.

The practitioner who extracts actionable insights, statistical models, and predictive algorithms from large data sets. This course sequences programming, database querying, and machine learning pipelines.

The data architect who designs, optimizes, and scales relational and non-relational database infrastructure. This course sequences relational SQL, Python data management, and backend query optimization.

The engineer who automates software delivery, infrastructure provision, and deployment pipelines. This course sequences automation scripting, server environments, security hardening, and cloud architecture.

The hardware-software engineer who develops firmware, driver interfaces, and micro-controller software. This course sequences low-level C++ and memory-safe Rust system programming.

The person who builds what a user actually touches. This course assumes you have never written code, and it is ordered the way the work is learned rather than the way it is advertised: the browser first, the language second, types third, the framework last. A framework learned before the language underneath it becomes a set of memorised incantations that break the moment something goes wrong. Each unit ends in something you build and can check yourself against, and the course ends in one build that needs all four units at once.

The engineer who bridges client interfaces and server architecture. This course assumes no background, and sequences the web from browser fundamentals to language, server runtime, database schemas, and modern UI component frameworks.

The software engineer who crafts interactive games, engine mechanics, and graphics pipelines. This course sequences low-level high-performance C++ and object-oriented C# gameplay logic.

The software engineer building enterprise backend services, microservices, and distributed applications. This course sequences core Java, relational SQL databases, and Node/API design principles.

The person who builds applications that run natively on phones and tablets. This course assumes no mobile experience, and it is ordered language foundation, type system, UI component architecture, then native mobile framework. Each unit ends in something you build and test on a device, and the course ends in a multi-screen cross-platform application.

The specialist who designs, configures, and maintains local and wide area network infrastructure. This course sequences networking fundamentals, enterprise routing, and perimeter security.

The engineer who designs, builds, and audits defensive systems and security controls. This course sequences network defense, vulnerability assessment, and Python automation.

The engineer who applies software engineering principles to infrastructure and operations problems to create highly scalable and reliable software systems.

The engineer who designs high-performance software, systems runtimes, and concurrent network services. This course sequences compiled systems programming, memory safety, and concurrent language design.

The core systems engineer building high-performance compilers, databases, operating systems, and distributed platforms using Rust, Go, and TypeScript tooling.

C# is Microsoft's general-purpose language, used for web services, desktop apps and most of Unity. This path covers the language and the .NET tooling around it.

C++ gives you direct control of memory and hides very little. It is harder than the alternatives and it is what games, browsers and trading systems are built in.

Go is deliberately small: you can hold the whole language in your head, and it was built for writing servers that handle a lot at once. Good second language, workable first one.

HTML is the structure of a page and CSS is how it looks. Neither is a programming language, and both are the fastest route to putting something you made in front of someone.

Java is verbose on purpose and runs almost everywhere. This path covers the language and the object model, which is the part that takes the longest to click.

JavaScript runs in every browser, so anything you write is something you can show someone. This path covers the language itself, then the browser APIs that make a page do something.

Node runs JavaScript on a server, which means the language you may already know arrives without a browser and with the filesystem, the network and a process to manage. This path starts at the runtime rather than at a framework, because almost everything that surprises people about Node is really about the event loop, and a framework only hides it.

Python reads close to plain English, which is why it is the usual first language. This path covers the syntax, then the standard library, then enough practice to write something of your own without a tutorial open beside you.

React is a library for building interfaces out of components. It assumes you can already write JavaScript, and most people who struggle with React are actually struggling with JavaScript.

React Native builds iOS and Android apps from React components. Learn React first: everything here assumes it, and the parts that are genuinely different are the platform APIs rather than the syntax.

Rust stops a whole class of memory bugs at compile time, and the price is a borrow checker that argues with you for the first few weeks. Everyone finds that part hard.

SQL is how you ask a database questions. It is small enough to learn quickly and deep enough that people use it for decades, and it barely changes between databases.

TypeScript is JavaScript with types checked before the code runs. This path assumes you can already write JavaScript, because TypeScript makes very little sense without it.

Arabic is written right to left, joins its letters, and leaves the short vowels out, so reading is a skill you build rather than a code you crack. This path teaches the script, then the three-letter root system that generates most of the vocabulary, then Modern Standard Arabic as it actually appears in writing and broadcast.

Danish grammar is close to Norwegian and Swedish, and its pronunciation is not: written Danish and spoken Danish diverge more than in any of its neighbours. This path faces that gap first, then the noun and verb system, then the vocabulary that everyday conversation runs on.

Dutch is close enough to English that vocabulary comes fast and far enough that the sounds and the word order do not. This path fixes the pronunciation English speakers guess wrong, then the rule that sends verbs to the end of a clause, then the thousand words that cover most of what gets said.

Finnish is spelled exactly as it sounds and stressed on the first syllable every time, so reading aloud is straightforward from day one. The difficulty is fifteen cases doing the work prepositions do in English, and a consonant change that fires whenever an ending is added. This path takes them in that order.

French writes far more letters than it says, and the gap between the page and the sound is where the difficulty sits. This path settles that first, then covers gender agreement and the two-verb past tense, then the everyday vocabulary that textbooks tend to skip.

German is spelled the way it is said, so reading aloud is possible on the first day, and then the grammar asks for four cases and a verb that does not stay where English puts it. This path takes them in that order and finishes on how compound words let you read vocabulary you were never taught.

The Greek alphabet is already half familiar from maths and science, which makes reading a short first step. After that Greek asks for four cases, gendered articles and an accent that has to sit on the right syllable. This path covers those, then the vocabulary that turns out to be shared with English.

Hebrew is written right to left and printed without vowels, so reading is the first real skill rather than a preliminary. Underneath that is a root and pattern system that makes vocabulary derivable rather than arbitrary. This path takes the script, then the roots, then modern spoken Hebrew.

Devanagari is a consistent script that pays back the week it takes, and Hindi grammar puts the verb at the end and uses postpositions where English uses prepositions. This path does the script, then that sentence structure, then the vocabulary Hindi shares with Urdu and English.

Italian spelling tells you almost everything about how a word sounds, including where the stress falls, which makes reading aloud straightforward from the start. The work is in agreement, articles that change shape depending on what follows them, and a past tense built from two verbs. This path takes those, then everyday vocabulary.

Japanese asks for two syllabaries before the grammar makes sense, and then rewards you with a sentence structure that is regular and a verb system with almost no irregular forms. This path does the scripts first, then particles, then kanji and counters as they come up in real use.

Hangul was designed to be learned quickly and it is, which means a beginner can be reading Korean within a day. After that the work is verb endings, which carry tense and politeness together, and two separate number systems. This path takes them in that order.

Latin is read rather than spoken, which changes what a course is for: the goal is getting through a text, not holding a conversation. This path settles pronunciation so you can read aloud, then works through the declensions and conjugations that carry all the meaning, then the vocabulary that survived into English.

Mandarin has no conjugation, no gender and no plurals, and puts all of its difficulty into tones and characters. This path handles pinyin and the four tones first, then the sentence frame and measure words, then the characters that repeat often enough to be worth learning early.

Norwegian has two official written standards and a spoken language that varies more by region than either of them suggests, and a grammar simpler than German or Russian. This path handles the two-standard question first, then the word order and endings, then everyday vocabulary. Bokmal leads, because that is what almost every course teaches.

Polish looks unpronounceable and is actually regular once the letter combinations are learned, and then asks for seven cases and a verb aspect choice in every sentence. This path settles the spelling first, then the grammar, then the vocabulary of shops, travel and small talk.

Portuguese looks like Spanish on the page and sounds nothing like it, and the nasal vowels are what separate the two. This path settles the sound first, then the verb system including the subjunctive that Portuguese uses more than English speakers expect, then everyday vocabulary. Brazilian pronunciation leads, with European material alongside it.

Cyrillic takes about a week and is the smallest obstacle in Russian. The real work is six cases, which change the ending of nearly every noun and adjective, and verb aspect, which makes you choose between two verbs for what English calls one. This path does the alphabet, then the cases, then the vocabulary that lives on top of them.

Spanish spelling maps to sound almost perfectly, so reading aloud comes fast and the real work is verbs. This path spends its first unit on the five vowels that never change, its second on the two verbs for "to be" and the tenses that follow, and its third on the vocabulary a conversation actually runs on.

Swedish grammar is simpler than its neighbours, with almost no verb conjugation, and puts its difficulty into pitch and into definite endings stuck on the back of the noun. This path covers those two, then the everyday vocabulary that makes conversation possible.

Tagalog is spelled the way it sounds and uses the Latin alphabet, so reading starts immediately. The grammar is the unfamiliar part: verbs change to mark which participant the sentence is about, rather than who is doing what. This path covers the sounds, then that focus system, then the mix of Tagalog and English that people actually speak.

Thai has no tenses, no plurals and no articles, and puts all of its difficulty into a 44-consonant script and five tones that the script encodes indirectly. This path takes the sounds and the writing first, then the markers that do the work tenses would, then everyday vocabulary.

Turkish is spelled exactly as it sounds and builds meaning by stacking suffixes on a stem in a fixed order, which makes it unusually regular once the system is clear. This path covers vowel harmony, then the suffix order, then everyday vocabulary.

Vietnamese is written in the Latin alphabet, which makes it look easy, and carries six tones, which is the part that actually decides whether you are understood. This path starts with the tones, moves to a grammar with no conjugation and no plurals, and ends on the classifiers and everyday words you need to say anything.

An agent is a model given tools and a loop: it decides what to call, reads the result, and decides again. That is genuinely useful and also the fastest way to build something expensive that does the wrong thing confidently. This path covers the loop, the tools, and how to keep it on a leash.

Understand model bias, prompt safety, hallucination mitigation, alignment, and governance.

Implement generative AI strategies, workflow automation, and ROI measurement in business.

Learn how to integrate OpenAI API endpoints, function calling, and Assistants API into web applications.

Claude is the model; Claude Code is the terminal tool that lets it read and edit a real codebase. This path starts with the interface you already have open, moves to skills and the file conventions that make Claude behave consistently, and ends with you having automated something you used to do by hand.

Explore image processing, object detection, segmentation, and convolutional neural networks.

Build neural networks from scratch and train models using PyTorch.

Start with what these models are and what they are not: text predictors trained at a scale that makes prediction look like understanding. Knowing where that breaks is most of what separates someone who uses them well from someone who gets burned by a confident wrong answer.

Explore the Hugging Face Hub, Transformers library, Datasets, and open-source model pipelines.

Frameworks for composing LLM chains, agent tools, document loaders, and memory buffers.

Learn how to fine-tune open-weight AI models using LoRA and PEFT techniques for specialized domains.

Run open-weight AI models (Llama 3, Mistral) locally on your hardware without third-party API keys.

Automate model training, tracking experiments, model registries, and REST API deployment.

Process images, diagrams, and text together using vision-language models (VLMs).

Master text preprocessing, tokenization, sentiment analysis, and sequence modelling.

Most prompt advice is a list of tricks with no way to tell whether any of them worked. This path teaches the parts that survive a model upgrade: being specific, giving examples, structuring the output you want, and measuring the result instead of trusting your impression of it.

Study Markov decision processes, Q-learning, and RLHF human preference tuning.

Retrieval-Augmented Generation combines vector search with large language models to ground AI responses in private or custom documents.

Master dense vector representations, distance metrics (cosine, dot product), and HNSW indexing.

Speech recognition (STT), speech synthesis (TTS), audio embeddings, and real-time voice agents.

A single-exam certification (SY0-701) covering security fundamentals, threats, architecture, operations, and program management.

A single-exam certification (N10-009) covering networking concepts, infrastructure, operations, security, and troubleshooting.

A two-exam certification (220-1201 and 220-1202) covering PC hardware, mobile devices, networking basics, operating systems, and troubleshooting for entry-level IT support roles.

A single-exam certification (CLF-C02) covering AWS cloud concepts, core services, security, and billing. The usual first AWS certification, with no technical prerequisites.

A single-exam certification (SAA-C03, the Associate level) on designing resilient, secure, cost-optimised architectures on AWS. Usually taken after Cloud Practitioner or equivalent hands-on experience.

A single-exam certification (AZ-900) covering cloud concepts and core Azure services, workloads, security, and pricing. The usual first Microsoft cloud certification.

A single-exam certification (200-301) covering network fundamentals, IP connectivity and services, security fundamentals, and automation. The standard entry networking certification.

A foundational certification (AIF-C01) covering AI, machine learning and generative AI on AWS, plus responsible AI and the security and governance around it. Written for people who use AI services rather than build them: AWS puts model coding, feature engineering and hyperparameter tuning explicitly out of scope.

AWS Certified CloudOps Engineer – Associate (SOA-C03) validates technical expertise in monitoring, reliability, deployment automation, security, and networking for AWS operations. Replaced the SysOps Administrator Associate exam (SOA-C02) on 30 September 2025.

AWS Certified Developer – Associate (DVA-C02) validates proficiency in developing, deploying, and debugging cloud applications using AWS services.

An associate certification (MLA-C01) about the engineering around a model rather than the maths inside it: preparing data, training and tuning, deploying to endpoints, orchestrating pipelines with CI/CD, and monitoring and securing what you shipped. AWS assumes about a year with SageMaker and a year in a related engineering role.

A single-exam certification (AI-901) covering AI concepts, responsible AI, and building generative AI, speech, vision and information-extraction solutions in Microsoft Foundry. It replaced AI-900 in April 2026 and is not the same exam: over half of it is implementation, and it expects you to read Python.

A single-exam certification (DP-900) covering core data concepts, relational and non-relational data on Azure, and what an analytics workload is made of. Describe-level throughout, and usually taken alongside or just after AZ-900.

EC-Council CEH (Certified Ethical Hacker v13) validates knowledge of ethical hacking tools, attack vectors, and security countermeasures.

CNCF CKA (Certified Kubernetes Administrator) is a performance-based practical exam that tests hands-on cluster installation, management, and troubleshooting.

CompTIA Cloud+ (CV0-004) certifies cloud engineers in multi-vendor cloud architecture, security, operations, and deployment.

CompTIA Cybersecurity Analyst (CySA+, CS0-004) certifies intermediate security analysts in incident response, threat detection, and vulnerability management. CS0-004 replaced CS0-003 in June 2026 and adds AI use in security operations.

CompTIA Linux+ (XK0-006) certifies system administrators in managing Linux operating systems, services and users, security, automation, and troubleshooting. XK0-006 replaced XK0-005, which retired in January 2026.

CompTIA PenTest+ (PT0-003) certifies ethical hackers and penetration testers in vulnerability scanning, exploitation techniques, and security reporting.

The three CompTIA exams people usually take together, in order: A+ (two exams) for IT support fundamentals, then Network+ for networking, then Security+ for security. Each one below starts with the official exam page, then a training course (free or paid), then practice material.

Google Cloud Associate Cloud Engineer certifies professionals who deploy applications, monitor operations, and manage enterprise GCP projects.

The HashiCorp Certified: Terraform Associate (004) exam verifies Infrastructure as Code (IaC) concepts, HCP Terraform, modules, state management, and HCL syntax.

The CISM certification establishes expertise in information security governance, risk management, program development, and incident response.

ISC2 CISSP (Certified Information Systems Security Professional) is the gold standard for security leaders, managers, and enterprise architects.

The ITIL 4 Foundation certification introduces the Service Value System (SVS), IT service management best practices, and Guiding Principles.

Azure Administrator Associate (AZ-104) certifies cloud administrators in managing Azure compute, storage, networking, identity, and governance.

The PL-300 exam validates data preparation, modeling, visualization, and report deployment using Microsoft Power BI.

A single-exam certification (SC-900) covering security, compliance and identity concepts and the Microsoft services behind them: Entra, Defender, Sentinel and Purview. It assumes no security background, and it is the usual first Microsoft security certification.

The OSCP is the industry standard hands-on penetration testing certification, evaluating active exploitation, privilege escalation, and active directory attacks.

PMI PMP (Project Management Professional) validates project leadership expertise across predictive, agile, and hybrid methodologies.