How ShareChat’s focus on AI and ML is helping the social networking site evolve and grow
Speaking at the 12th edition of YourStory’s flagship startup tech conference, TechSparks 2021, ShareChat’s Hastagiri Vanchinathan offers his take on ‘Growing with AI: The ShareChat and Moj Story’.
ShareChat made headlines in early 2021 by raising $502 million to build India’s largest AI-powered content ecosystem. Artificial Intelligence (AI) has been an integral part of the company, with the technology often being the company’s guiding light.
Speaking at TechSparks 2021, YourStory’s flagship startup-tech conference, Hastagiri Vanchinathan, Director, AI - Feed Ranking, ShareChat, revealed the ‘Growing with AI: The ShareChat and Moj Story’. The session, hosted by Chintan Shah, Head - Digital Native and Software Business, Google Cloud, focused on the role technology played in the company.
Hastagiri said the company’s vision is to build the largest AI-powered content platform in India. AI plays a key role in all areas of operations, but it doesn’t stop there. From AI to video technologies, cloud platforms, and large-scale data processing, all new-age technologies have been implemented for the growth of ShareChat and Moj, which have slowly become the largest regional social media platforms in India.
While ShareChat has over 180 million Monthly Active Users, Moj has over 160 million Monthly Active Users.
“If you look at India’s population and the amount of content consumption and creation, you’ll realise why we need technology,” Hastagiri said.
![..](https://images.yourstory.com/cs/2/2d86ed30b28211e8b2e7114aea10c711/Imageqtzo-1635314441660.jpg?fm=png&auto=format)
Growing with AI
AI is critical to the company’s growth, and plays an active role in various areas.
“From user acquisition to choosing UI elements and the central core where we do feed ranking and monetisation, everything is powered by AI,” said Hastagiri, adding that what will be a good journey for a user on their first day on the app and predicting what decisions to take are also results derived from AI and data-backed planning.
Hastagiri said their teams are organised into five different verticals:
- Feed ranking
- Monetization AI
- Content understanding
- Camera AI
- AI Infra
“Apart from this, in every operation area of the company, data analytics and data-driven decision-making plays a key role,” he said.
Acing time reduction and faster experimentation
Hastagiri said time reduction and learning from data are two extremely important things. To that end, their partnership with Google Cloud has helped them with some major changes. When it comes to feed ranking, the team has used GPUs better for both training and serving feed content. This hasn’t just reduced their training time, but also lowered costs.
“More recently, our move to TPU saw us cut our training time by 2-3X and it allowed us faster experimentation,” Hastagiri added.
He revealed the team has hundreds of machine learning models in production. “In order to orchestrate this army of models, the Compose service was very useful for us. Similarly, we have an enormous amount of data. We relied on Google Cloud to pick them up for analytics and ML tasks.”
The Moj and ShareChat teams have been achieving these milestones even during the pandemic. The Moj app has millions of active users today, but was built in less than 30 hours a year ago.
“The AI/ML team contributed heavily to the growth of Moj. The ML team was 15 people over a year and half ago, and today, we've grown to 90 people,” Hastagiri said.
To log in to our virtual events platform and experience TechSparks 2021 with thousands of other startup-tech enthusiasts from around the world, join here. Don't forget to tag #TechSparks2021 when you share your experience, learnings and favourite moments from TechSparks 2021.
For a line-up of all the action-packed sessions at YourStory's flagship startup-tech conference, check out TechSparks 2021 website.
![Gif](https://images.yourstory.com/cs/2/2d86ed30-b282-11e8-b2e7-114aea10c711/1635258277736.gif?fm=png&auto=format)
Edited by Teja Lele