Enterprise AI Architect
A structured path from LLM fundamentals through production agent systems, security, platforms, and enterprise architecture.
Understand Models
Build a working mental model of tokens, transformers, attention, embeddings, inference, and sampling.
- LLM mental model
- Tokens
- Context
- Transformers
- Attention
- Embeddings
- Inference
- Sampling
- Reasoning models
How LLMs Work — Complete Guide
End-to-end interactive visual overview of the entire LLM pipeline on one page.
Open →Tokenization
BPE, SentencePiece, vocabulary tradeoffs, and how text becomes numbers.
Open →Transformers
The transformer architecture, positional encodings, and scaling laws that power modern AI.
Open →Attention
Query, Key, Value mechanics — Flash Attention, GQA, MQA, and multi-head patterns.
Open →Embeddings
Similarity, RAG, vector search, and how meaning is represented geometrically.
Open →Inference Pipeline
Prefill/decode phases, KV-cache, speculative decoding, and serving optimization.
Open →Probability & Sampling
Softmax, temperature, beam search, top-k/top-p, and calibration methods.
Open →Build AI Applications
Move from model literacy to application architecture: context, retrieval, tools, and basic evaluation.
- Structured outputs
- Prompt architecture
- Context engineering
- Model selection
- Retrieval
- RAG
- Tool calling
- Basic evaluation
Build Agent Systems
Decide when an agent is warranted, then structure loops, memory, orchestration, and human approval.
- Agents vs workflows
- Agent loops
- Planning
- Tool execution
- Memory
- State
- Orchestration
- Human approval
- Multi-agent systems
Building Enterprise AI Agents
Production harness: runtime, orchestration, tools, memory, RAG, security, observability, and production controls.
Open →Trip Planner Agent
A practical multi-agent walkthrough — planning, execution, and coordination patterns.
Open →Context Engineering
17 prompt and context engineering techniques for building reliable AI systems.
Open →Understand Agent Runtimes
Treat the harness as the boundary: execution, checkpoints, queues, retries, sandboxing, and policy.
- Agent Harness
- Runtime boundaries
- Execution
- Checkpointing
- Durable workflows
- Queues
- Retries
- Idempotency
- Sandboxing
- Policy enforcement
Secure Agent Systems
Give agents identity, constrain tools, and authorize every hop — not just the first request.
- Agent identity
- Delegation
- Authorization
- Prompt injection
- Capability boundaries
- Tool permissions
- Credential management
- Runtime authorization
- Auditability
Operate AI in Production
Evals, tracing, cost, latency, routing, SLOs, and incident handling for systems that are allowed to act.
- Evals
- Tracing
- Observability
- Reliability
- Failure recovery
- Cost
- Latency
- Model routing
- SLOs
- Incident handling
Architect Enterprise AI
Platforms, gateways, integration, multi-tenancy, governance, and build-versus-buy at company scale.
- AI platforms
- Model gateways
- Enterprise integration
- Distributed systems
- Multi-tenancy
- Governance
- Control planes
- Build vs buy
- Enterprise reference architecture