AI Spaceship

For developers & engineersLearn AI from first principles. Apply it to real problems.Hands-on courses that take you from the math and mechanics behind AI systems to production-ready implementations. No black boxes — you build every layer yourself, then apply what you learn to solve actual business problems.Executable Python labs in the browser·Incremental, validation-driven exercises·Built for engineers who shipStart learning CoursesEach course is a self-contained path from concept to implementation. Work through theory, write code, and validate your understanding at every step.Build Your Own LLM6 modules · 18 lessonsGo from raw text to a working language model. Covers data pipelines, tokenization, transformer architecture, training loops, and post-training — all with hands-on Python labs.Open course Retrieval-Augmented Generation5 modules · 14 lessonsBuild production RAG systems: chunking strategies, embedding models, vector stores, retrieval pipelines, and evaluation frameworks for grounded AI applications.Open course AI Agents & Tool Use4 modules · 13 lessonsDesign and implement autonomous agents with tool calling, planning loops, memory systems, and reliable execution — patterns you can ship to production.Open course Evaluation & Testing for AI Systems5 modules · 17 lessonsBuild systematic evaluation pipelines: define metrics, create test suites, run automated evals, and set up CI guardrails for AI-powered features.Open course --- CourseLLM Engineering From First PrinciplesModule 1: IntroductionThe LLM PipelineNumPy for Neural NetworksVectors and SimilarityModule 2: Data PreparationReading Input DataBPE Tokenization with tiktokenData Sampling with Sliding WindowToken and Positional EmbeddingsBatching with a DataLoaderModule 3: Attention MechanismsDot-Product SimilaritySelf-Attention from ScratchScaled Dot-Product AttentionCausal (Masked) AttentionMulti-Head AttentionModule 4: The Transformer BlockGELU Activation FunctionPosition-Wise Feed-Forward NetworkLayer NormalizationThe Transformer BlockStacking Blocks into a GPTModule 5: Training the GPTCross-Entropy LossLoss GradientsBackpropagation Through a Linear LayerGradient DescentThe Training LoopAutoregressive Text GenerationModule 6: Inference and EvaluationSaving and Loading WeightsPerplexityEnd-to-End Inference PipelineKV Cache for Fast InferenceBeam SearchModule 7: Fine-Tuning for ClassificationTransfer LearningReplacing the Output HeadFreezing LayersClassification Training LoopEvaluation MetricsModule 8: Instruction Fine-TuningInstruction Data FormatChat TemplatesSupervised Fine-TuningResponse ExtractionInstruction-Following EvaluationLLM Engineering From First Principles8 modulesBuild Your Own LLMLearn the full LLM pipeline through compact lessons and executable labs.Build, train, and adapt GPT-style language models through structured hands-on lessons.Course SnapshotModules8Lessons39Completion0%Sign in to track progressModule 1: Introduction3 lessonsThe LLM PipelineOpenLesson in Module 1: IntroductionOpen lessonNumPy for Neural Networks LockedLesson in Module 1: IntroductionLockedVectors and Similarity LockedLesson in Module 1: IntroductionLockedModule 2: Data Preparation5 lessonsReading Input Data LockedLesson in Module 2: Data PreparationLockedBPE Tokenization with tiktoken LockedLesson in Module 2: Data PreparationLockedData Sampling with Sliding Window LockedLesson in Module 2: Data PreparationLockedToken and Positional Embeddings LockedLesson in Module 2: Data PreparationLockedBatching with a DataLoader LockedLesson in Module 2: Data PreparationLockedModule 3: Attention Mechanisms5 lessonsDot-Product Similarity LockedLesson in Module 3: Attention MechanismsLockedSelf-Attention from Scratch LockedLesson in Module 3: Attention MechanismsLockedScaled Dot-Product Attention LockedLesson in Module 3: Attention MechanismsLockedCausal (Masked) Attention LockedLesson in Module 3: Attention MechanismsLockedMulti-Head Attention LockedLesson in Module 3: Attention MechanismsLockedModule 4: The Transformer Block5 lessonsGELU Activation Function LockedLesson in Module 4: The Transformer BlockLockedPosition-Wise Feed-Forward Network LockedLesson in Module 4: The Transformer BlockLockedLayer Normalization LockedLesson in Module 4: The Transformer BlockLockedThe Transformer Block LockedLesson in Module 4: The Transformer BlockLockedStacking Blocks into a GPT LockedLesson in Module 4: The Transformer BlockLockedModule 5: Training the GPT6 lessonsCross-Entropy Loss LockedLesson in Module 5: Training the GPTLockedLoss Gradients LockedLesson in Module 5: Training the GPTLockedBackpropagation Through a Linear Layer LockedLesson in Module 5: Training the GPTLockedGradient Descent LockedLesson in Module 5: Training the GPTLockedThe Training Loop LockedLesson in Module 5: Training the GPTLockedAutoregressive Text Generation LockedLesson in Module 5: Training the GPTLockedModule 6: Inference and Evaluation5 lessonsSaving and Loading Weights LockedLesson in Module 6: Inference and EvaluationLockedPerplexity LockedLesson in Module 6: Inference and EvaluationLockedEnd-to-End Inference Pipeline LockedLesson in Module 6: Inference and EvaluationLockedKV Cache for Fast Inference LockedLesson in Module 6: Inference and EvaluationLockedBeam Search LockedLesson in Module 6: Inference and EvaluationLockedModule 7: Fine-Tuning for Classification5 lessonsTransfer Learning LockedLesson in Module 7: Fine-Tuning for ClassificationLockedReplacing the Output Head LockedLesson in Module 7: Fine-Tuning for ClassificationLockedFreezing Layers LockedLesson in Module 7: Fine-Tuning for ClassificationLockedClassification Training Loop LockedLesson in Module 7: Fine-Tuning for ClassificationLockedEvaluation Metrics LockedLesson in Module 7: Fine-Tuning for ClassificationLockedModule 8: Instruction Fine-Tuning5 lessonsInstruction Data Format LockedLesson in Module 8: Instruction Fine-TuningLockedChat Templates LockedLesson in Module 8: Instruction Fine-TuningLockedSupervised Fine-Tuning LockedLesson in Module 8: Instruction Fine-TuningLockedResponse Extraction LockedLesson in Module 8: Instruction Fine-TuningLockedInstruction-Following Evaluation LockedLesson in Module 8: Instruction Fine-TuningLocked --- CourseRetrieval-Augmented GenerationModule 1: Foundations — What RAG Is and Why It MattersThe RAG PipelineText Splitting BasicsRepresenting Text as NumbersModule 2: Embeddings and Similarity SearchWord Vectors with NumPyBuilding a Vector IndexApproximate Nearest NeighborsModule 3: Retrieval PipelineEnd-to-End RetrievalBM25 and Keyword SearchHybrid SearchModule 4: Generation and Prompt EngineeringPrompt ConstructionCalling an LLM APIGrounded GenerationModule 5: Production PatternsEvaluation FrameworkChunking Strategies and TuningRetrieval-Augmented Generation5 modulesBuild Your Own LLMLearn the full LLM pipeline through compact lessons and executable labs.Build production RAG systems from scratch: chunking, embeddings, vector search, retrieval pipelines, and evaluation — all with hands-on Python labs.Course SnapshotModules5Lessons14Completion0%Sign in to track progressModule 1: Foundations — What RAG Is and Why It Matters3 lessonsThe RAG PipelineOpenLesson in Module 1: Foundations — What RAG Is and Why It MattersOpen lessonText Splitting Basics LockedLesson in Module 1: Foundations — What RAG Is and Why It MattersLockedRepresenting Text as Numbers LockedLesson in Module 1: Foundations — What RAG Is and Why It MattersLockedModule 2: Embeddings and Similarity Search3 lessonsWord Vectors with NumPy LockedLesson in Module 2: Embeddings and Similarity SearchLockedBuilding a Vector Index LockedLesson in Module 2: Embeddings and Similarity SearchLockedApproximate Nearest Neighbors LockedLesson in Module 2: Embeddings and Similarity SearchLockedModule 3: Retrieval Pipeline3 lessonsEnd-to-End Retrieval LockedLesson in Module 3: Retrieval PipelineLockedBM25 and Keyword Search LockedLesson in Module 3: Retrieval PipelineLockedHybrid Search LockedLesson in Module 3: Retrieval PipelineLockedModule 4: Generation and Prompt Engineering3 lessonsPrompt Construction LockedLesson in Module 4: Generation and Prompt EngineeringLockedCalling an LLM API LockedLesson in Module 4: Generation and Prompt EngineeringLockedGrounded Generation LockedLesson in Module 4: Generation and Prompt EngineeringLockedModule 5: Production Patterns2 lessonsEvaluation Framework LockedLesson in Module 5: Production PatternsLockedChunking Strategies and Tuning LockedLesson in Module 5: Production PatternsLocked --- CourseAI Agents & Tool UseModule 1: Foundations — The Agent LoopWhat Is an Agent?The ReAct LoopStructured Output ParsingModule 2: Tool UseDefining ToolsTool DispatchMulti-Tool AgentsTool Error HandlingModule 3: Planning & MemoryPlanning with SubgoalsConversation MemorySummarized MemoryModule 4: Multi-Agent SystemsAgent HandoffOrchestrator PatternGuardrails and StoppingAI Agents & Tool Use4 modulesBuild Your Own LLMLearn the full LLM pipeline through compact lessons and executable labs.Design and implement autonomous agents with tool calling, planning loops, memory systems, and reliable execution — patterns you can ship to production.Course SnapshotModules4Lessons13Completion0%Sign in to track progressModule 1: Foundations — The Agent Loop3 lessonsWhat Is an Agent?OpenLesson in Module 1: Foundations — The Agent LoopOpen lessonThe ReAct Loop LockedLesson in Module 1: Foundations — The Agent LoopLockedStructured Output Parsing LockedLesson in Module 1: Foundations — The Agent LoopLockedModule 2: Tool Use4 lessonsDefining Tools LockedLesson in Module 2: Tool UseLockedTool Dispatch LockedLesson in Module 2: Tool UseLockedMulti-Tool Agents LockedLesson in Module 2: Tool UseLockedTool Error Handling LockedLesson in Module 2: Tool UseLockedModule 3: Planning & Memory3 lessonsPlanning with Subgoals LockedLesson in Module 3: Planning & MemoryLockedConversation Memory LockedLesson in Module 3: Planning & MemoryLockedSummarized Memory LockedLesson in Module 3: Planning & MemoryLockedModule 4: Multi-Agent Systems3 lessonsAgent Handoff LockedLesson in Module 4: Multi-Agent SystemsLockedOrchestrator Pattern LockedLesson in Module 4: Multi-Agent SystemsLockedGuardrails and Stopping LockedLesson in Module 4: Multi-Agent SystemsLocked