AI Engineering with Python & LangChain

Description

This comprehensive program equips engineers with the skills to design, build, evaluate, and deploy production-grade AI applications. Covering LLM orchestration, RAG architectures, agentic systems, model evaluation, fine-tuning, and cloud deployment, the course bridges experimentation and enterprise implementation. Participants gain practical expertise in transforming large language models into reliable, scalable, and maintainable AI systems.

Indicative Duration: 88 training hours
*Duration is adjusted based on the final scope and the target audience.

Scope

1. Python Foundations for AI Engineering	1.1 Environment Setup	Setting up Virtual Environments (venvs) Create python projects using uv Managing dependencies Exploring libraries
	1.2 Structuring Code	Functions and Classes Strong Typed extensions Pydantic & @dataclasses
	1.3 Logging and Testing	Implementing basic and configured logging Unit and Integration testing
2. LLM Application Development Foundations	2.1 Introduction to LangChain	Exploring different models & clients (OpenAI, AzureOpenAI, Groq, Local) Connecting to models (local & cloud) .env file configuration Introduction to LangChain Expression Language (LCEL) Building a simple chain with Prompts and Models
	2.2 Working with Documents: Loaders & Splitters	Loading content from Text/PDF sources Understanding LangChain Document structure Text splitters and splitting strategies
	2.3 Embeddings & Vector Stores	Deep dive into embeddings Creating embeddings
	2.4 Building a RAG System with Chroma	Understanding embeddings in depth Using ChromaDB as local vector store Creating and storing embeddings from documents
	2.5 Advanced Retrieval Strategies	Beyond similarity search MMR & Self-Query retrievers Building advanced RAG chains
3. Agentic Systems with LangGraph	3.1 Introduction to LangGraph	State, Nodes, Edges Building a linear graph Conditional logic Graph with AI integration
	3.2 Implementing Agentic Tools	Defining tools with @tool decorator LLM tool calling mechanics Testing tools in isolation
	3.3 Managing State & Memory	Conversational loops Using graph state Re-entrancy concepts
	3.4 Persistence with Checkpointers	Persistent state importance LangGraph Checkpointers Using MemorySaver
	3.5 Composable Agents	Agents as tools concept Designing worker agents Packaging agents as tools
	3.6 Recursive & Hierarchical Agents	Manager agent delegation Recursive workflows Problem decomposition
	3.7 Asynchronous Flow for Performance	Sync vs Async concepts Async methods in LangChain Async LangGraph execution
4. Observability & Knowledge Systems	4.1 Introduction to LangSmith	Overview of LangSmith platform
	4.2 Introduction to Knowledge Graphs	KG concepts (Nodes, Relationships) Neo4j & Cypher Populating a simple KG
	4.3 LangChain & Knowledge Graph Integration	Text-to-Cypher KG-RAG vs Vector-RAG KG-based QA system
5. Application Interfaces	5.1 Streamlit Core Concepts	Building UI for LangChain Interactive chat layouts
	5.2 Conversational Interfaces with Chainlit	Chat-first framework Layout & element management Chainlit vs Streamlit comparison
	5.3 Advanced UI & Feedback Loop	User feedback mechanisms Capturing feedback Handling file uploads for RAG
	5.4 Interactive Chat Layout	Persistent chat with gr.ChatInterface Connecting UI to LangGraph backend Managing state in UI
6. Protocols, Advanced Tooling & Model Control	6.1 Introduction to MCP	MCP architecture Consuming MCP in LangChain Building first MCP server
	6.2 Advanced Agent Tools	Structured tool wrappers Async tools Error handling
	6.3 Tool-Based Modality Switching	Multi-modal agent design Router node in LangGraph Modality switching logic
	6.4 Advanced Prompt Engineering	OpenAI parameters (temperature, top_p) System roles Chain-of-Thought patterns
	6.5 Managing Complex Model I/O	Consistent prompt formats Output-to-input chaining Output parsing & validation
	6.6 Handling Multimodal Inputs	Vision, audio models Processing image, PDF, audio inputs
7. Evaluation & Model Optimization	7.1 Evaluation Theory & Dataset Creation	LLM evaluation metrics Designing evaluation datasets
	7.2 Practical Evaluation with LangChain	Evaluation chains Automated evaluation pipeline Result analysis
	7.3 Additional Evaluation Strategies	Hallucination testing Anti-hallucination strategies
8. Model Ecosystem & Fine-Tuning	8.1 The World of Hugging Face	HF Hub & libraries Speech-to-Text use cases
	8.2 Fine-Tuning	Why & when to fine-tune Preparing datasets
	8.3 Model Training Loop	Trainer API setup Training configuration Saving model artifacts
	8.4 Evaluating Fine-Tuned Models	Qualitative vs Quantitative evaluation BLEU/ROUGE metrics Human-in-the-loop evaluation
	8.5 Integrating Small & Fine-Tuned Models	HF Inference API LangChain wrapper SLM in LangGraph workflow
9. Cloud Deployment & MLOps in Azure	9.1 Azure Functions	Serverless concepts HTTP Trigger endpoint Additional triggers & data binding Deploying LangChain as web API
	9.2 Azure AI Search	From keywords to vectors Search architecture Index, indexer, pipeline
	9.3 Professional Project Setup & MLOps	Project goal tracking Git workflow CI/CD with GitHub Actions
	9.4 Introduction to MCP & Azure ML	AML Workspace setup Managing data & model assets
	9.5 Production Pipelines & Compliance	Azure-ready ML pipelines with MCP Responsible AI compliance Monitoring deployed pipelines
	9.6 Azure AI Studio & Prompt Flow	Platform overview Building visual Prompt Flows Integrating LLMs & tools
	9.7 Advanced Prompt Flow	Python tools in flows A/B testing variants Dynamic inputs
	9.8 Deployment & Evaluation with Prompt Flow	Built-in evaluation tools Model hosting options Managed endpoint deployment

Learning Objectives

Upon completion of the course participants will be able to:

Design and implement end-to-end LLM-powered applications using LangChain and LangGraph.
Build advanced Retrieval-Augmented Generation (RAG) systems with vector databases and knowledge graphs
Engineer stateful, tool-enabled, multi-agent architectures with orchestration logic
Apply advanced prompt engineering, model control, and multimodal processing techniques
Evaluate, benchmark, and improve model reliability using structured evaluation pipelines
Fine-tune and integrate open-source models into production workflows
Deploy AI applications using cloud-native architectures and MLOps best practices

Target Audience

Roles: AI Engineers, Software Engineers, Solution Architects, Technical Leads
Seniority: Junior to Senior Professionals

Prerequisite Knowledge

Solid Python foundations
Beneficial but not mandatory
- Understanding of APIs and RESTful services
- Familiarity with JSON and structured data handling
- Basic knowledge of machine learning and LLM concepts
- Experience with Git and software development workflows
- Cloud fundamentals

Delivery Method

Sessions can be delivered via the following formats:

Live Online – Interactive virtual sessions via video conferencing
On-Site – At your organization’s premises
In-Person – At Code.Hub’s training center
Hybrid – A combination of online and in-person sessions

The training methodology combines presentations, live demonstrations, hands-on exercises and interactive discussions to ensure participants actively practice AI in realistic work scenarios.

Tags: AI Engineering

Date

On Demand

Organizer

Code.Hub

Email

[email protected]