by Aabir Abubaker Kar
~1 min read


  • posts
  • projects
  • Projects


  • deep learning timeseries classification nlp

Quick summary:

  • What I did:
    • co-founded, an AI company building tools to assist ordinary newsreaders in fact-checking and media curation
    • user research and market research, including interviews with journalists, designers, and potential users
    • built mock-ups (using Figma!) of the proposed interface
    • built a Python-based MVP using natural language processing, machine learning and network analysis
    • scoped cloud and software requirements for building a web-app and GUI around the MVP
    • applied to Y Combinator in Summer 2020, just before the pandemic hit
  • Things I learnt:
    • how to ruthlessly prioritize feature-development from a business perspective
    • how to build product-market fit using market and user research
    • how to conceptualize and iterate on a machine learning product that isn’t yet clearly defined
    • how to scope a machine learning project with variable scope
    • how to leverage cloud technologies for flexible scaling requirements
  • When was this?
    • January 2020 through September 2020

The year is 2020 and the reliability of things you see online has never been lower. Rumors, clickbait, biased language, and straight up lies are abundantly found on the internet, and no Google search is immune. The 24-hour news cycle fills your feed with junk designed to provoke and rile you, hoping to stay your eyeballs, pause your scrolling, and monetize your attention.

Our information diets are mostly junk, and it’s common to feel powerless about it. Regina Catipon and I founded to fix that.

Regina had been a journalist who saw what social media did to the news industry. I was a machine learning expert with a keen interest in digital media and disinformation. We started in January 2020, with discussions about our interests and the drive to create real-world impact. We ramped up to an ambitious Y Combinator Application in March, following which the pandemic hit us all. tried to stay afloat thereafter, iterating and working on an MVP through September, even hiring our first engineer (the brilliant and industrious Adarsh Mathew) - but our spatial and temporal separation and academic commitments (all of us had to start writing our theses in September 2020) forced us to sunset the project.

Unless the world magically fixes this problem, we hope to come back to it at a later date!