Skip to content
አማርኛEnglish

Theme

ይህ ገጽ ገና ወደ አማርኛ አልተተረጎመም፤ ስለዚህ በእንግሊዝኛ ቀርቧል። የእንግሊዝኛውን ገጽ ክፈቱ

በቅርቡ

AI with Python

Move past toy prompts to build production-ready AI applications: structured outputs, semantic search, RAG, and autonomous tool-using agents.

  • 12 ክፍሎች
  • 12 ትምህርቶች
  • Shallows to Landfall
  • Free to learn

This course is being written. Its lessons will be free here, like every lesson on Mawj. The outline below is what it will teach; follow the Telegram channel to hear the day it opens.

Get notified on Telegram

What you'll be able to do

  • Call leading LLM APIs programmatically and reliably
  • Extract validated data using Pydantic structured outputs
  • Implement semantic search and retrieval-augmented generation (RAG)
  • Build autonomous tool-using AI agents that solve complex workflows

The full syllabus

12 modules, from the Shallows down. Open any module to see its lessons.

The Shallows

Understanding LLMs, calling APIs from Python, and prompt engineering.

  1. , coming soon
    1How LLMs workIntuition for tokens, context windows, and probabilistic text generation.
    1. 1.1Tokens and context windowsበቅርቡ
  2. , coming soon
    2Calling LLM APIsConnect to model APIs from Python, manage keys, and stream responses.
    1. 2.1Your first LLM API callበቅርቡ
  3. , coming soon
    3Prompt engineeringInstructions, few-shot examples, system messages, and role definitions.
    1. 3.1Writing effective promptsበቅርቡ

Currents

Structured outputs, data manipulation, embeddings, and RAG.

  1. , coming soon
    4Structured outputs with PydanticEnforce strict JSON schemas and parse LLM outputs into typed objects.
    1. 4.1Extracting structured dataበቅርቡ
  2. , coming soon
    5Data essentials: NumPy & pandasJust enough data handling for vectors, matrices, and tabular datasets.
    1. 5.1Working with datasetsበቅርቡ
  3. , coming soon
    6Embeddings and semantic searchConvert text into high-dimensional vectors and perform cosine similarity search.
    1. 6.1Generating text embeddingsበቅርቡ
  4. , coming soon
    7Retrieval-augmented generation (RAG)Chunk documents, retrieve relevant contexts, and synthesize grounded answers.
    1. 7.1Building a RAG pipelineበቅርቡ

Deep Water

Tool calling, autonomous agents, evaluation, and production safety.

  1. , coming soon
    8Tool use and agentsLet an LLM call Python functions, search databases, and take actions.
    1. 8.1Function calling and tool useበቅርቡ
  2. , coming soon
    9Evaluating AI featuresMeasure accuracy, hallucinations, and regression across prompt revisions.
    1. 9.1Automated AI evaluationsበቅርቡ
  3. , coming soon
    10Cost, latency, and safetyRate limiting, caching, cost monitoring, and guardrails for production.
    1. 10.1Production guardrails and costበቅርቡ

Landfall

Ship AI behind FastAPI and complete the document assistant capstone.

  1. , coming soon
    11Shipping AI behind FastAPIExpose streaming AI features and RAG pipelines via fast web endpoints.
    1. 11.1Streaming API endpointsበቅርቡ
  2. , coming soon
    12Capstone: Document assistantBuild an end-to-end assistant that answers questions over custom PDF and markdown files.
    1. 12.1Building the document assistant (ፕሮጀክት)በቅርቡ

Who it's for

  • Python developers wanting to integrate modern AI into real products
  • Software engineers who want to build with RAG, embeddings, and agents
  • Students aiming for practical AI engineering skills beyond theory

What you need

  • Comfortable with Python functions, dictionaries, and virtual environments
  • An internet connection and an API key for experimentation