Develop Generative AI Applications: Get Started

🎓 Part of the IBM RAG and Agentic AI Professional Certificate #

Author: Ahmad Tawil
Date Completed: October 12, 2025
Certificate: Coursera Verification Link →

---

📘 Course Overview #

This repository documents my complete work and labs for the IBM course “Develop Generative AI Applications: Get Started”, offered through Coursera.
The course provides hands-on experience in Generative AI, Prompt Engineering, and LangChain frameworks, culminating in a Flask-based AI application powered by IBM watsonx.ai models.

💡 Key Learning Outcomes #

• Understand Generative AI foundations, prompt engineering, and in-context learning
• Master LangChain components (LLMs, Prompts, Chains, Agents, LCEL)
• Build composable AI workflows using LCEL pipelines
• Develop and deploy a Flask Generative AI web app integrating LangChain + Watsonx models
• Compare LLMs (Llama 3, Granite) and implement structured JSON output parsing

---

🧩 Course Modules #

Module 1 – Foundations of Generative AI and Prompt Engineering #

Explored the core principles of generative AI and how prompt engineering improves LLM performance.
Implemented LangChain Prompt Templates and practiced in-context learning (zero-shot, one-shot, few-shot).

Skills & Tools

• Prompt Engineering 🧠
• NLP Concepts (Tokenization, NER, POS)
• LangChain PromptTemplate
• Python + Jupyter Notebooks

Lab: In-Context Learning and Prompt Templates for Advanced AI

---

Module 2 – Introduction to LangChain in Generative AI Applications #

Learned the modular structure of LangChain — LLMs, Chat Models, Prompts, Output Parsers — and built sequential chains, memory, and agents.
Adopted the LangChain Expression Language (LCEL) for clear and composable AI pipelines.

Key Concepts

Feature	Description
Chains	Sequential workflows where output → next input
Memory	Maintains context for multi-turn conversations
Agents	Dynamically choose tools/APIs to execute tasks
LCEL Syntax	Pipe operator (`

Lab: Build Smarter AI Apps – Empower LLMs with LangChain

---

Module 3 – Build a Generative AI Application with LangChain #

Developed a full-stack Flask Generative AI application integrating IBM watsonx.ai models (Llama 3, Granite) with LangChain.

Tech Stack

Category	Tools
AI Framework	LangChain + LangChain-IBM
Models	Meta Llama 3 / IBM Granite
Backend	Flask (API + UI)
Parsing	Pydantic + JsonOutputParser
Language	Python 3.11
Frontend	HTML + Flask Templates

Project Highlights

• Integrated LangChain chains into Flask backend
• Defined Pydantic schemas for validated JSON outputs
• Compared Llama 3 vs Granite in runtime and response style
• Produced structured LLM responses (summary, sentiment, reply)

Example Output

{
  "summary": "Large Language Models (LLMs) are AI algorithms that process and generate human-like language.",
  "sentiment": 100,
  "response": "They are trained on vast amounts of text data to learn patterns and relationships in language."
}

Screenshot

---

🧠 Course Architecture Summary #

User (UI)
 │
 └──> Flask (app.py)
      │ receives user prompt + model
      └──> model.py → ChatWatsonx chain
             │ PromptTemplate + JsonOutputParser
             └──> Watsonx.ai Model (Llama3 / Granite)
                     │ returns structured JSON
                     └──> Flask renders output to UI

---

🏅 Skills Demonstrated #

• Generative AI Concepts & Foundation Models
• Prompt Engineering & In-Context Learning
• LangChain Framework (Chains, Memory, Agents, LCEL)
• Flask Web App Development for AI Applications
• IBM Watsonx Model Integration (Llama 3, Granite)
• Structured Output Parsing with Pydantic & JSON

---

📚 Tools and Technologies #

Python 3.11 · LangChain · Flask · IBM Watsonx.ai · Pydantic · JsonOutputParser · LLMs (Llama 3, Granite) · HTML/CSS · VS Code · Jupyter Lab

---

🧩 Future Enhancements #

• Add RAG (vector store integration) for contextual retrieval
• Extend JSON schema with tone/confidence fields
• Implement multi-turn memory and session storage
• Deploy to IBM Cloud or Vercel for public access

---

🏁 Conclusion #

Successfully completed all three modules of IBM’s Develop Generative AI Applications: Get Started course.
This project demonstrates the full journey — from mastering prompt engineering to building a production-ready Flask Generative AI web app powered by LangChain and IBM Watsonx foundation models.

🧩 Ready to move on to advanced RAG and Agentic AI development.

---

✅ Verified Certificate: Coursera.org/verify/X6ECLAMRH2YQ