VEL ORENDISVel Orendis

LLM applications, end to end

From a model’s first answer to retrieval, agents, evaluation and monitoring: the whole path, with the background each part needs.

54 lessons, about 9.6 h. Anything you already know, you can prove and skip.

0 of 54 lessonsquests defended, 2 open now

Start with What the Model Cannot Know
  1. What the Model Cannot KnowWorking with LLMs, 8 min
    Open
  2. Calling a model: messages, tokens and costWorking with LLMs, 10 min
    Opens later
  3. Prompts that ground answersWorking with LLMs, 10 min
    Opens later
  4. The RAG pipelineSearch and RAG, 10 min
    Opens later
  5. The request path and where time goesFoundations, 10 min
    Opens later
  6. Testing strategyShipping, 10 min
    Opens later
  7. Configuration, secrets and containersShipping, 10 min
    Opens later
  8. Lexical search and BM25Search and RAG, 12 min
    Opens later
  9. Ingestion apart from queryingSearch and RAG, 10 min
    Opens later
  10. The Stale CacheCaching, 12 min
    Open
  11. Cache patterns and where caches liveCaching, 10 min
    Opens later
  12. HTTP, CDN and prompt cachingCaching, 12 min
    Opens later
  13. From notebook to serviceShipping, 10 min
    Opens later
  14. Structured output and validationWorking with LLMs, 10 min
    Opens later
  15. Function callingAgents, 12 min
    Opens later
  16. The agent loop and its budgetsAgents, 12 min
    Opens later
  17. When not to use an agentAgents, 8 min
    Opens later
  18. Safeguards: guardrails, injection, permissionsAgents, 12 min
    Opens later
  19. Frameworks and what they hideAgents, 10 min
    Opens later
  20. Embeddings and similaritySearch and RAG, 12 min
    Opens later
  21. Vector search, exact and approximateSearch and RAG, 12 min
    Opens later
  22. Tables, keys and modellingDatabases, 10 min
    Opens later
  23. Indexes and query plansDatabases, 12 min
    Opens later
  24. Vectors in PostgreSQLSearch and RAG, 10 min
    Opens later
  25. Shipping embeddingsSearch and RAG, 10 min
    Opens later
  26. Ground truth for searchRetrieval quality, 10 min
    Opens later
  27. Hit rate, MRR and other metricsRetrieval quality, 12 min
    Opens later
  28. Hybrid search and rank fusionRetrieval quality, 12 min
    Opens later
  29. APIs: resources, idempotency, paginationQueues and streams, 10 min
    Opens later
  30. Throughput, utilisation and queueingFoundations, 10 min
    Opens later
  31. Background jobs, retries and rate limitsQueues and streams, 10 min
    Opens later
  32. Multi-agent systems and orchestrationAgents, 12 min
    Opens later
  33. Context engineeringWorking with LLMs, 10 min
    Opens later
  34. AI-assisted engineeringShipping, 12 min
    Opens later
  35. Freshness, metadata and live retrievalSearch and RAG, 10 min
    Opens later
  36. Transactions and the write-ahead logDatabases, 10 min
    Opens later
  37. CI/CD, migrations and rollbacksShipping, 12 min
    Opens later
  38. Progressive delivery and version skewShipping, 10 min
    Opens later
  39. Logging every LLM callEvaluation and monitoring, 10 min
    Opens later
  40. Observability: logs, metrics, tracesReliability, 12 min
    Opens later
  41. Dashboards and traces for LLM appsEvaluation and monitoring, 12 min
    Opens later
  42. Offline and online evaluationEvaluation and monitoring, 8 min
    Opens later
  43. LLM as a judgeEvaluation and monitoring, 12 min
    Opens later
  44. Evaluating agentsAgents, 12 min
    Opens later
  45. Shipping an LLM featureShipping, 12 min
    Opens later
  46. Tuning search without fooling yourselfRetrieval quality, 10 min
    Opens later
  47. Feedback and online judgesEvaluation and monitoring, 10 min
    Opens later
  48. Query rewriting, thresholds and the honest no-answerRetrieval quality, 10 min
    Opens later
  49. Chunking long documentsSearch and RAG, 12 min
    Opens later
  50. RerankingRetrieval quality, 10 min
    Opens later
  51. Sampling and choosing a modelWorking with LLMs, 10 min
    Opens later
  52. Comparing prompts and modelsEvaluation and monitoring, 10 min
    Opens later
  53. Ingestion pipelinesEvaluation and monitoring, 12 min
    Opens later
  54. MCP: tools over a protocolAgents, 12 min
    Opens later