AI Ecosystem
The real providers, models, and tools behind the AI concepts in this curriculum — how they layer together into a production stack.
The Layered Stack
Explore by Category
AI Application Development
Turn an existing application into an AI-enabled one — frontend, backend, and everything in between.
8 topicsData Engineering & GenAI
Move, process, and prepare the data that feeds AI pipelines.
7 topicsAI Data & Retrieval
The real-world implementation of embeddings, vector search, and RAG in production.
7 topicsAgents & AI Integration
The real-world frameworks and operational concerns behind production agent systems.
8 topicsCloud & AI Infrastructure
Where AI workloads actually run — cloud platforms, containers, and GPU infrastructure.
8 topicsProduction AI / LLMOps
Operate AI systems reliably once they’re live — evaluation, observability, cost, and deployment.
8 topicsAI Security & Governance
Build AI systems that protect data, models, users, and enterprise workflows.
10 topicsReal-World AI Stack
Who and what actually implements the concepts you’ve learned — providers, models, and tools.
6 topicsTechnology Reference
Every provider, model family, and tool referenced across the curriculum’s Real-World Stack panels.
Providers
OpenAI
A model lab best known for the GPT family, offered through its own API and through Azure OpenAI.
Anthropic
A model lab focused on AI safety research, best known for the Claude family.
Google
Both a model lab (the Gemini family) and, through Google Cloud, a major cloud platform for hosting and serving AI workloads.
Microsoft
Operates Azure, including Azure OpenAI Service, and ships AI assistants across its own product line under the Copilot name.
Meta
Publishes the Llama family as openly-licensed weights that can be self-hosted rather than accessed only through a hosted API.
Mistral AI
A European model lab publishing both open-weight and hosted models under the Mistral family name.
Cohere
A model provider focused on enterprise use cases, particularly retrieval, embeddings, and reranking.
AI21 Labs
A model lab known for the Jamba family, which mixes transformer and state-space architecture components.
Alibaba Cloud
Publishes the Qwen family of open-weight models alongside its own cloud infrastructure business.
DeepSeek
A model lab publishing the openly-licensed DeepSeek family, notable for competitive performance at lower training cost.
Amazon
Operates AWS, including the Bedrock service for accessing multiple model families, and publishes its own Nova model family.
NVIDIA
The dominant GPU vendor for AI training and inference, and publisher of AI infrastructure software such as NeMo.
Model Families
GPT
OpenAI's general-purpose LLM lineage, offered through the OpenAI API and Azure OpenAI.
Claude
Anthropic's model family, commonly used for long-context reasoning, coding, and agentic tool use.
Gemini
Google's multimodal model family, integrated across Google Cloud's Vertex AI platform.
Llama
An openly-licensed model family that can be self-hosted, fine-tuned, or served through many third-party platforms.
Qwen
An openly-licensed model family from Alibaba Cloud, spanning a wide range of model sizes.
Mistral
Mistral AI's model family, spanning both open-weight and hosted-only models.
DeepSeek
An openly-licensed model family known for strong reasoning and coding performance relative to training cost.
Amazon Nova
Amazon's own model family, accessed through AWS Bedrock alongside models from other providers.
Frameworks
LangChain
A library of composable building blocks for LLM applications — prompts, chains, memory, and tool integrations.
LangGraph
A library for building agents and multi-step workflows as an explicit graph of nodes and edges, including branching and loops.
LlamaIndex
A framework focused on connecting LLMs to external data — ingestion, indexing, and retrieval pipelines.
Semantic Kernel
Microsoft's open-source SDK for integrating LLMs into applications, with first-class .NET, Python, and Java support.
CrewAI
A framework for orchestrating multiple role-based agents that collaborate on a shared task.
Vector Databases
Pinecone
A fully-managed vector database built specifically for similarity search at scale.
Qdrant
An open-source vector database with filtering, hybrid search, and both self-hosted and managed options.
Weaviate
An open-source vector database with built-in hybrid search and modules for generating embeddings inline.
Milvus
An open-source vector database designed for very large-scale similarity search workloads.
Chroma
A lightweight, developer-friendly embedding database often used for prototyping and smaller applications.
pgvector
A PostgreSQL extension that adds vector similarity search directly inside a regular Postgres database.
OpenSearch
An open-source search engine (a fork of Elasticsearch) with vector search capability alongside traditional full-text search.
Elasticsearch
A widely-used search and analytics engine that also supports vector similarity search.
MongoDB Atlas Vector Search
A vector search capability built into MongoDB's managed Atlas platform.
AI Gateways
LiteLLM
An open-source library and proxy that exposes many model providers through one OpenAI-compatible interface.
OpenRouter
A hosted routing service that provides unified API access to many model providers.
Portkey
An AI gateway offering routing, caching, fallback, and observability for LLM calls.
Cloudflare AI Gateway
A gateway built into Cloudflare's edge network for caching, rate limiting, and logging LLM requests.
Observability
LangSmith
LangChain's platform for tracing, debugging, and evaluating LLM application runs.
Langfuse
An open-source platform for tracing, evaluation, and cost/latency monitoring of LLM applications.
Arize Phoenix
An open-source LLM observability and evaluation tool, including embedding and retrieval visualization.
OpenTelemetry
A vendor-neutral standard and toolset for collecting traces, metrics, and logs across distributed systems.
Datadog
A general-purpose monitoring and observability platform with dedicated LLM monitoring features.
Helicone
An open-source logging and analytics layer that sits between an application and its model provider.
Evaluation
Ragas
An open-source library specifically for evaluating RAG pipelines — retrieval relevance and answer faithfulness.
DeepEval
An open-source testing framework for LLM applications, styled like a unit-testing library.
Security & Guardrails
NVIDIA NeMo Guardrails
An open-source toolkit for adding programmable guardrails around LLM conversations.
Guardrails AI
An open-source library for validating and correcting LLM inputs and outputs against defined rules.
Microsoft Presidio
An open-source library for detecting and redacting personally identifiable information (PII) in text.
Cloud Platforms
AWS
Amazon's cloud platform, including Bedrock for model access and SageMaker for custom model hosting.
Azure
Microsoft's cloud platform, including Azure OpenAI Service and Azure AI Foundry.
Google Cloud
Google's cloud platform, including Vertex AI for building and serving models.
Infrastructure
Kubernetes
An open-source system for deploying, scaling, and managing containerized applications.
Amazon EKS
AWS's managed Kubernetes service.
Azure AKS
Azure's managed Kubernetes service.
Google GKE
Google Cloud's managed Kubernetes service.
Docker
A tool for packaging an application and its dependencies into a portable container image.
Data
Apache Kafka
A distributed event-streaming platform for publishing and consuming continuous streams of records.
Apache Spark
A distributed data-processing engine for large-scale batch and streaming workloads.
PostgreSQL
A widely-used open-source relational database.
Snowflake
A managed cloud data warehouse for large-scale analytics workloads.
Databricks
A unified data and AI platform combining data engineering, analytics, and machine learning workflows.