← Noesa

Using AI, day by day

For Backend developers moving into a manager/integrator role — practical fluency to use and direct AI, not train models. · 17 days · Tech

A 17-day course
One short lesson a day · go at your own pace
Log in to start
The days
Listen to the whole course · 87 min
  1. 0Set up your AI benchBy the end you can open Ari's scratch project and get one real model reply from Claude Code. Listen · 5 min
  2. 1See what an LLM predictsBy the end you can explain an LLM as a next-token predictor and connect that mechanism to both fluency and fabrication. Listen · 5 min
  3. 2Count the tokensBy the end you can estimate a request's token budget and explain how input and output tokens affect cost, speed, and context. Listen · 5 min
  4. 3Turn meaning into vectorsBy the end you can compute cosine similarity over fixed embeddings and use the ranking to choose a relevant Ari passage. Listen · 5 min
  5. 4Trace the assistant layersBy the end you can distinguish a base model, post-training, system instructions, and the product runtime around Claude. Listen · 5 min
  6. 5Steer output and verify limitsBy the end you can write a bounded prompt contract and add verification for unsupported claims and nondeterministic output. Listen · 5 min
  7. 6Run a model locallyBy the end you can run one open-weight model with Ollama and decide whether local execution fits a workload. Listen · 5 min
  8. 7Ground answers with RAGBy the end you can assemble and evaluate a retrieve-then-answer flow over Ari's runbooks. Listen · 5 min
  9. 8Let the model request toolsBy the end you can implement the request-tool-result loop and keep execution authority outside the model. Listen · 5 min
  10. 9Connect capabilities with MCPBy the end you can explain MCP's host-client-server architecture and choose between local stdio and remote Streamable HTTP transport. Listen · 5 min
  11. 10Package capability as skillsBy the end you can package Ari's incident-triage instructions and assets as a reusable, reviewable skill. Listen · 5 min
  12. 11Build an explicit agent loopBy the end you can describe and implement a bounded agent loop with tools, state, and a stop condition. Listen · 5 min
  13. 12Choose workflows and guard agentsBy the end you can choose a fixed workflow or agent and specify guardrails against prompt injection and unsafe actions. Listen · 5 min
  14. 13Map Google's agent stackBy the end you can place Gemini, Vertex AI, ADK, and Agent Engine into model, development, and managed-runtime layers for a use case. Listen · 5 min
  15. 14Direct coding agents wellBy the end you can give a coding agent a scoped task, review its evidence, and decide whether the change is ready. Listen · 5 min
  16. 15Map the AI tool landscapeBy the end you can classify an unfamiliar AI tool by layer, contract, state, authority, and evaluation burden. Listen · 5 min
  17. 16Scope Ari for productionBy the end you can produce a build-ready AI feature plan that chooses prompt, RAG, workflow, or agent and defines cost, controls, and review. Listen · 5 min