AI for Education
Global HackWeek
July 28 - August 11

Event over, results announced. $19K in prizes!

Challenge 1: Government of India Data Science Challenge

BONUS PRIZE: Opportunity to present solution to NITI Aayog, Government of India, and CEOs of top Indian startups.

Analyze large education datasets from government/public schools to figure out which factors most influence student performance, enrollment, dropout, or transition from primary to secondary. Based on your analysis, suggest where the government should be investing more or less, and how.

Use the government-collected datasets attached/linked below for your analysis or obtain your own dataset from another source (e.g. education non-profits, kaggle, google, web scraping public web locations, etc.)

What you will submit:

  • Excel file, Jupyter notebook, Matlab/R output of your analysis
  • Blog post or PDF with your commentary on the analysis, and recommendations for government

Suggested Education Datasets (India-specific)

Challenge 2: Government of India AI Challenge

BONUS PRIZE: Opportunity to present solution to NITI Aayog, Government of India, and CEOs of top Indian startups.

Recent advances in natural language processing and machine comprehension have focused almost exclusively on English. For a linguistically diverse nation like India, scalable AI solutions for education will require language models for Hindi and other Indian languages.

In this AI challenge, apply NLP / ML techniques to building systems that comprehend, correct, or respond to text in Hindi or other Indian languages.

The models and algorithms you write could enable government and the ed-tech community to build chatbots for information retrieval and language learning, to translate scanned government documents from English to local languages, to build searchable databases of non-English books / documents, etc.

Examples of problems you can work on:

  • Grammar checking and language scoring for the chosen (lots of tools / tutorials for this on our resources page)
  • Intent Detection for questions in the chosen language
  • Entity Extraction for questions in the chosen language (see here for an English example from IBM Watson)
  • Word2Vec (vector representation of text) for the chosen language
  • Neural Machine Translation for English to Indian languages (e.g. using this open-source tool from Harvard NLP, or with a custom sequence-to-sequence algorithm, see tutorials like this one)

Suggested Datasets

Challenge 3, 4, 5: Suggested by IBM Watson/Research team

Query classification and Content Recommendation:

Build a chatbot which can understand user’s query/question either in the form of text or voice and suggest the topics and comments from stack overflow dataset.  To do this, one needs to process the stack overflow dataset and cluster it based on topics and its popularity. Similarly, user query needs to be processed and identify the topics on which chatbot can suggest the content. For example, user query can be “What are the differences between functional programming language and object-oriented programming?” and chatbot can suggest the topics related to “Lips, Scheme, C++, JAVA, Julia”.

Design the application to navigate around the topic discussed in the order of topic difficulty and popularity using the stack overflow dataset.

Hint: Use Text to Speech or Speech to Text for intuitive interaction. Use Watson conversation service for chatbot. Use NLU Watson APIs for topic understanding.

Link Topics to Relevant News Stories:

Using the CNN news dataset and education text books, link text book topics with news stories of relevance towards the experiential learning.

By understanding the text book topics with their related new stories helps in enriching the user learning experience.  For example, while learning the text book topic such as “The Science of Earthquakes - the basics in brief” and it’s one of the related news such as 5.8 magnitude earthquake hits western Montana can be retrieved from CNN news dataset and recommended to the user. To do this, one can process the news articles and text books, and identify the news that can be mapped to the text book topics.

Hint: Use NLU APIs for topic identification in news dataset and text book, and use Rank and Retrieve for semantic search

Automatic Article Generation:

Generate articles by augmenting text with relevant images from open source using WikiQA dataset.

Hint: Use Watson visual recognition service for image understanding and NLU APIs for topic understanding

WildCard Challenge

Apply your knowledge of CS, Data Science, Machine Learning, and NLP to build open-source AI for education, solving any problem in education you're passionate about, e.g. personalization of content and delivery, intelligent assessments, instant question-answering, clustering and classification of student learning styles, etc.

Your outputs can be any of the following:

  • usable products (e.g. chatbots, web apps, mobile apps)
  • data sets (e.g. scrape public sources to collect data useful for future AI work in education)
  • tutorials / how-to articles for AI-education work you’ve previously done

Prizes (for all challenges)




  • Sponsored By
  • Omidyar Network



  • Sponsored By
  • IBM


in GCP Credits

  • Sponsored By
  • Google Developer Groups


in AWS Credits

  • Sponsored By
  • Amazon Web Services

Free Cloud Credits for All Participants
We've run out! Sorry


in Watson access (6 months)

  • Sponsored By
  • IBM


in GCP Credits

  • Sponsored By
  • Google Developer Groups


in AWS credits

  • Sponsored By
  • AWS Educate


in Digital Ocean credits

  • Sponsored By
  • Digital Ocean

HackWeek Schedule (July 28 - Aug 11)

Friday, July 28

5:00pm IST (7:30am EST)

Resources/Datasets released, Hacking period starts!

Friday, July 28

5:30pm IST - 6:15pm IST

Registration, Set up Credits, Networking

Venue: IIT Delhi, Lecture Hall Complex
Register in-person if you haven't already, get help setting up free credits, and meet each other.

Friday, July 28

6:15pm IST - 6:30pm IST

Introduction + HackWeek Instructions

Venue: IIT Delhi, LH121
We'll cover important details such as challenge problems, judging criteria, prizes, credits, submission instructions, and agenda for the weekend and the week.

Friday, July 28

6:30pm IST - 7:30pm IST

Opening Session: Unsolved Problems and Startup Opportunities in Education AI

Vishal Dixit, Sarvesh Kanodia (Omidyar Network)

Venue: IIT Delhi, LH121

Omdiyar Network has invested in startups like, Khan Academy, AltSchool, Teach for India, and AspiringMinds. Vishal and Sarvesh from the Omidyar team will present the landscape of problems and opportunities for Education entrepreneurs today.

Friday, July 28

7:30pm - 8:30pm IST

Idea Pitches for Team Formation (Dinner provided)

Venue: IIT Delhi, LH121

Looking for teammates? Pitch yourself, your background and what you're looking for (2 minutes per person), your teammates will come find you.

Saturday, July 29

10:30am - 11:00am IST

Breakfast with IBM

Venue: IIT Delhi, Lobby of Lecture Hall Complex

Saturday, July 29

11:00am - 1:00pm

IBM Watson Hands-on Workshop

Danish Contractor and Karan Chaturvedi (IBM)

Venue: IIT Delhi, LH121

Danish, an NLP expert from IBM Research, will spend the first 45 minutes covering opportunities for AI in Education, and Karan will spend the next hour walking through how to build an end-to-end AI-powered web app using Watson's javascript API.

Saturday, July 29

1:00pm - 1:30pm IST

Lunch with Google Developers

Venue: IIT Delhi, Lobby of Lecture Hall Complex

Saturday, July 29

1:30pm - 5:00pm IST

End-to-End ML with Google TensorFlow

Manoranjan Padhy and Rohit Gupta (Google)

Venue: IIT Delhi, LH121

TensorFlow is a free and powerful platform that makes machine learning easy. Rohit and Manoranjan will walk through a quick intro to ML with TensorFlow, run an Ideation session on how to build a successful model, and then demo a model in TensorFlow using Google Cloud DataLab.

Saturday, July 29

5:00pm - 5:30pm IST

KEYNOTE: A Vision for Indian Education through AI and Open Data

Mr. Amitabh Kant (CEO, NITI Aayog, government of India), introduced by Prof. Rao (Director, IIT Delhi)

Venue: IIT Delhi, LH121

Mr. Kant will talk about the urgent need for change in Indian education (with a view towards job creation and employability), the role of AI in the government’s digital roadmap, and the importance of open data in government and education. He will conclude by presenting Challenge Problems to participants for HackWeek and beyond.

Saturday, July 29

5:30pm - 6:30pm IST

Dinner (full meal) and Team Time

Venue: IIT Delhi, LH121 and Lobby

Sunday, July 30

12:00pm - 1:00pm IST


Venue: IIT Delhi, Lobby of Lecture Hall Complex

Sunday, July 30

1:00pm - 3:00pm

How to win a Data Science Contest: Extracting insights from big data in Education

Dipyaman "Deep" Sanyal, CFA

Venue: IIT Delhi, LH121

A Kaggle-style in-class contest run by Deep, a Program Chair for Analytics at the Bridge School of Management. In this hands-on workshop, Deep will lead participants through the phases of data exploration, feature engineering and predictive modelling using econometrics, ML, and an education-specific dataset, ending with a contest for the best participant-created model.

Monday July 31 - Friday Aug 11

All day / every day

Hack from Anywhere!

Friday, Aug 11

11:59pm IST (2:29pm EST)

Submissions Close, Judging Starts

Tuesday, Aug 15

5pm IST (7:30am EST)

Results Announced

Advisors and Judges

Hal Abelson

CS Professor at MIT. Father of open-source: Founding Director of Creative Commons, Public Knowledge, and Free Software Foundation.

Alexander M. Rush

Asst. Professor at Harvard School of Engineering. Runs Harvard NLP. Researching neural machine translation and text summarization.

Denny Britz

Deep Learning @ Google Brain, Stanford CS, writes at Researching conversational AI and re-inforcement learning.

François Chollet

Creator of, a widely used Python deep learning library; AI & Deep Learning researcher at Google; Global rank 17 on Kaggle.

Joel Tetreault

Director of Research @ Grammarly. Previously NLP @ Yahoo Labs. Essay/language-scoring expert. Runs NLP for Education workshop.


Anshul Bhagi

Founder @ Camp K12

Computer Science undergrad and masters from MIT, MBA from Harvard. ex-Google, ex-Apple. Currently bringing coding to k-12 education in India, excited about AI tutors and decentralized, p2p education.

Nikhila Ravi

Developer @ Facebook

Studied engineering at Cambridge University, and Machine Learning / Data Science at Harvard as a Kennedy Scholar. Avid cricketer for MCC women, vegan food blogger at

Shreyas Deshmukh

PM @ Microsoft

Computer Science graduate from VIT, previously mentored ed-tech startup Sling (acquired) in India. Building an AI-powered motion/joint tracking platform for video input.

Partners and Sponsors

Media and Press

Outreach Partners


If you’re a student (college/high school) or a working professional passionate about AI and eager to solve problems in the field of education, you’ve come to the right place. Anyone, anywhere in the world can take part for free.
Of course! You can participate as an individual or in teams of up to 4 people. Each person on your team has to register separately.
You can work on any problem in the domain of education that you're passionate about, e.g. personalization of content and delivery, intelligent assessments, instant question-answering, clustering and classification of student learning styles, question intent-detection, etc. We’ve got some free resources at for you to get started but, feel free to use any other resource available to you.
Absolutely. Choose whatever platform or tech stack you prefer the most.
You need to upload your submission (whether it's code, a dataset, a how-to article) on GitHub, provide a short description and we highly recommend creating a 2 min video demo about your project too. We will share the submission form and other details during the hackweek.
Submissions are due at midnight (IST) on August 11th. From 12th August, we'll be working with our superstar judging panel on evaluating submissions. The results will be announced on August 15th, 2017.
Your submissions will be evaluated by our panel of ML / NLP experts on the following criteria:
  • Importance of problem - what is the impact of the problem for education?
  • Project idea - creativity of proposed solution and scope
  • Implementation and quality of final product/output - e.g. how well does the product work? Is the scraped dataset well structured?
  • Technical design and architecture - e.g choice of dataset, features, API, ML model, Neural Network design
  • Visual design and Usability - quality of UX and UI
Please email us at or reach us on Slack if you’re in any way confused or concerned! We’d love to help you out :)