AI_&_Dev_DevClub

🤖 CAIC Summer of Technology 2025: ML+Dev Track

Introduction

Welcome to the Machine Learning + Development (ML+Dev) track of CAIC Summer of Technology 2025!

This vertical is designed not only to teach you the basics of ML but to help you build, integrate, and deploy full systems that solve real-world problems. You’ll go from a raw dataset to a deployed web app that predicts and generates marketing content — a hands-on journey through ML, backend, and frontend integration.

---

Why ML+Dev?

This track emphasizes practical product building using ML. Here's what you’ll gain:

• Full-stack product thinking: Connect data science to real-world use cases.
• ML as a component: Learn to embed ML models in real software systems.
• Deployment and DevOps: Deploy your own models online with interactive UIs.
• Resume-worthy work: Have a hosted link, codebase, and demo ready by the end.

---

What Problem Are We Solving?

Inspired by Adobe Experience Cloud, you’ll build a content intelligence tool with two tasks:

Task 1: Predict Likes (Behavior Simulation)

Given a tweet's metadata, predict how many likes it will receive. Marketers can use this to estimate engagement before posting.

Task 2: Generate Tweet Text (Content Simulation)

Given metadata, generate tweet content that is engaging and aligned with the brand tone.

Together, these simulate and create content — powered by ML but designed as usable tools.

---

Weekly Roadmap — CAIC Summer of Tech: ML+Dev Track

Week 1: Problem Understanding & Dataset Familiarization

Focus on:

• Understanding the ML problem
• Loading & inspecting data
• Light preprocessing and EDA
• Planning integration (what features will go into the model? how will API accept inputs?)

✅ Deliverable: Cleaned dataset, insight-driven features, and structured Colab notebook

---

Week 2: Build Like Prediction Engine (Task 1)

Focus on:

• Feature engineering
• ML model development for likes
• Save model using joblib or pickle
• Create an API that takes input and returns predicted likes

✅ Deliverable: Trained model + REST API (Flask/FastAPI)

---

Week 3: Build Tweet Generation Engine (Task 2)

Focus on:

• Fine-tune or use a text generation model
• Wrap it into a callable function
• Serve it as an API

✅ Deliverable: Working generative model + API to return generated tweet

---

Week 4: Frontend + System Integration

Focus on:

• Create UI using Streamlit, React or Flask
• Connect frontend → backend APIs → ML models
• Make it interactive (input metadata → get likes and/or tweet)

✅ Deliverable: Locally working end-to-end app

---

Week 5: Deployment & Showcasing

Focus on:

• Containerization (optional)
• Deploy to Hugging Face Spaces, Render, Vercel, etc.
• Make a project video
• Write a devlog/blog

✅ Deliverable: Hosted working demo + GitHub repo + presentation

---

Tools & Technologies

• ML: Scikit-learn, Transformers, XGBoost, HuggingFace
• Dev: Python, FastAPI / Flask, Streamlit / React
• Deployment: HuggingFace Spaces, Render, Docker
• Data Analysis: Pandas, Seaborn, Matplotlib

---

Learning Resources

• Kaggle: Intro to ML
• ML Model Deployment Guide
• FastAPI Docs
• Streamlit Docs
• HuggingFace Spaces Guide

---

FAQ

Q: Will I learn ML from scratch?
A: Yes — but even more, you’ll learn how to use it in real apps.

Q: Is this dev or ML-focused?
A: It's ML-powered, but product/dev-focused — your main goal is to build something usable with ML inside it.

---

Let’s Go!

Build it. Use it. Ship it. 🚀