phonlamaiphoto - stock.adobe.com

Ola’s Krutrim builds ‘AI-first’ sovereign cloud for India

Krutrim is building a vertically integrated technology stack to make AI affordable, scalable and sovereign for Indian businesses while catering to the country’s linguistic needs

Krutrim, the artificial intelligence (AI) venture from Indian ride-hailing and electric vehicle (EV) giant Ola, is building a vertically integrated sovereign cloud and AI stack for India, using Cloudera's data platform as a core component of its architecture.

Speaking at Cloudera’s regional customer conference in Singapore, Navendu Agarwal, senior vice-president and head of business at Krutrim, said the company aims to create a hyperscale cloud designed from the ground up for India’s specific economic, linguistic, and data sovereignty needs, moving away from what it calls “copying the Western architecture”.

As group CIO for parent company Ola, Agarwal also oversees a business that spans ride-hailing, financial services and the largest electric vehicle manufacturing operation in India, all of which generate a staggering amount of data.

“We produce petabytes of data every day from different kinds of geospatial data, consumer insights and manufacturing sensors,” Agarwal said. That’s in addition to the data generated every five seconds by over a million scooters on the road.

This data, ranging from battery performance and sensor logs to rider behaviour patterns, allows Ola to perform critical correlations, such as linking a vehicle incident back to a specific manufacturing fault or a battery’s charge-discharge cycle.

To address the complexity of Ola’s data landscape, Krutrim has had to build a new kind of data platform designed to serve machine learning models rather than business intelligence dashboards. “We’re not designing for the dashboard; we’re designing for AI models and the future of AI,” Agarwal said.

More importantly, the platform will help to tackle challenges specific to India, which uses more than 22 languages, numerous dialects and code-mixed languages such as “Hinglish”. The country also has vast quantities of unstructured documents such as handwritten forms and scanned PDFs.

Beyond technical hurdles, Krutrim’s mission is driven by the need for more affordable AI services and digital sovereignty. Agarwal pointed out the prohibitive cost of using Western hyperscalers, whose pricing for graphics processing units (GPUs) is not viable for the cost-conscious Indian market.

“We need to fundamentally think how we can make AI affordable, scalable and sovereign,” he said, adding that geopolitical tensions can limit access to Western technology and affect critical infrastructure such as oil plants.

To build its sovereign cloud and AI stack, Krutrim has turned to Cloudera to provide the underlying technology for its data platform, allowing it to focus on creating the vertically integrated, AI-first experience.

Formally announced last week, the partnership will see Cloudera’s data platform power large-scale analytics and data lake workloads for Ola on the Krutrim Cloud, with plans to offer the solution to other enterprise customers in India soon.

The ultimate vision for Krutrim is an AI-powered cloud where developers and businesses can build and deploy complex applications using simple, natural language prompts, abstracting away the underlying complexity of data connectors, security and infrastructure.

For example, a user could simply ask the platform’s Kruti AI assistant to create a plant management website, connect it to the data lake, generate dashboards from camera feeds and schedule a daily report. “That’s the future and where we are going,” Agarwal said.

Read more about AI in India

  • IT professionals in India need to unlearn old skills and embrace AI, automation and specialised domains as the country cements its position as a global tech hub.
  • Modular datacentre infrastructure can help Indian enterprises build faster, greener and more flexible datacentres to cope with the demand for local data storage and growing use of AI and edge computing.
  • Jio Platforms, AMD, Cisco and Nokia have teamed up to develop an Open Telecom AI Platform aimed at improving the efficiency and security of telco networks through AI and automation.
  • India has earmarked significant investments in AI, global capability centres and digital public infrastructure while boosting training and education in its latest budget.

Read more on Artificial intelligence, automation and robotics