Skip to content
ISDIC

Community curriculum · Community 02

AI & ML

Move students from "I've used ChatGPT" to "I built and deployed a model." Every session involves running actual code in Google Colab — no theory-only lectures. The goal is a portfolio of real ML projects recruiters can see and test.

This is the planned activity framework for a full academic year. It describes what each chapter is set up to run, not a record of events already delivered.

All four domains and the annual calendar

activities / year
14
bootcamps
2
projects built
4+
code sessions
Biweekly

The year, quarter by quarter

11 planned entries across the academic year.

  1. Jan — Mar

    1. Python & data bootcamp

      2-day intensive

    2. EDA with Indian datasets

      Workshop

    3. First Kaggle team entry

      Competition

  2. Apr — Jun

    1. Supervised learning workshop

      Workshop

    2. Build-a-model Fridays ×3

      Recurring session

    3. LLM awareness session

      Workshop

  3. Jul — Sep

    1. Deep learning bootcamp

      2-day intensive

    2. RAG & LangChain workshop

      Workshop

    3. Paper reading club ×2

      Research session

  4. Oct — Dec

    1. AI for social good sprint

      6-week project

    2. Demo day & awards

      Showcase

Flagship event

Flagship — Twice/year

Weekend ML Bootcamp

From zero to deployed model in 2 days. Day 1: Python, NumPy, Pandas, data cleaning. Day 2: Scikit-learn, model training, evaluation, and Streamlit deployment on Hugging Face Spaces. Students leave with a live app and a shareable link.

  • Sentiment analyzer
  • Price predictor
  • Image classifier
  • Recommendation engine
  • Custom chatbot

Activities

Monthly sessions

  • Build-a-model Fridays

    Biweekly

    A community lead builds a model live from scratch every other Friday. Audience codes along in Colab. Topics rotate across classification, NLP, computer vision, and time series.

    • Google Colab
    • Live coding
    • Project-based

  • LLM & generative AI series

    Monthly

    Monthly deep-dives into LLMs, prompt engineering, RAG, and fine-tuning. Build a chatbot on a custom dataset, compare API providers, or fine-tune a model on free Colab GPU.

    • Prompt engineering
    • RAG
    • LangChain
    • Fine-tuning

  • Research paper reading club

    Monthly

    One member presents a landmark ML paper in plain English. Papers: Attention Is All You Need, ResNet, CLIP, Stable Diffusion, AlphaFold. Builds technical vocabulary that separates good engineers from great ones.

    • Transformers
    • CNNs
    • Diffusion models

  • Data analytics workshop

    Quarterly

    Real Indian datasets — Census, Zomato, IPL, Nifty stocks. Students explore, visualize, and present findings. Teaches EDA, data storytelling, and dashboarding.

    • Pandas
    • Matplotlib
    • Tableau Public
    • Power BI

Competitions and projects

  • Kaggle competition sprint

    Quarterly

    Community teams participate in a live Kaggle competition together. Senior members mentor beginners. The target is a leaderboard ranking — showing companies a Kaggle profile is proof of real ability.

    • Kaggle
    • Feature engineering
    • Ensemble methods

  • AI for social good challenge

    Annual

    Teams build ML solutions for real Indian problems — crop yield, disease detection, traffic management. 6-week mentor-guided sprint. Winning project gets presented to a company partner and may receive seed funding.

    • Social impact
    • Team project
    • Demo day

Skill progression

  1. Level 01

    Foundation

    Months 1 – 3

    • Python for data science
    • NumPy and Pandas
    • Data visualization basics
    • First Kaggle submission

  2. Level 02

    ML practitioner

    Months 4 – 8

    • Supervised and unsupervised ML
    • Model evaluation and tuning
    • Neural networks intro
    • Deployed Streamlit app

  3. Level 03

    Advanced

    Months 9 – 12

    • Deep learning with PyTorch
    • LLM fine-tuning and RAG
    • MLOps fundamentals
    • Kaggle top 30% ranking

Tools used

Target outcomes