 "Satinder Pal Singh, head of solution architecture at AWS India")
With India aiming to become a $1 trillion digital economy by end of this year, cloud infrastructure and generative AI are becoming increasingly central to that vision. Amazon Web Services (AWS), the first global hyperscaler to be empanelled by the Ministry of Electronics and Information Technology (MeitY), has positioned itself as a key player in this transformation. In a phone interview with Peerzada Abrar, Satinder Pal Singh, head of solution architecture at AWS India and South Asia, discusses how AWS is enabling generative AI adoption across industries, including startups, supporting language diversity, promoting responsible AI, and helping bridge the nation’s digital skills gap. Edited excerpts:
AWS became the first hyperscaler to be empanelled by MeitY (Ministry of Electronics and Information Technology). How has this enabled you to accelerate generative AI adoption across key sectors in India?
We announced our first AWS region in India in Mumbai in 2016. In 2017, we became the first global cloud service provider to be empanelled by MeitY, laying the foundation for organisations in India to adopt secure and scalable cloud services. This has accelerated generative AI adoption across sectors such as healthcare, education, and financial services. With Amazon Bedrock now generally available in the AWS Asia Pacific (Mumbai) Region, institutions can securely access foundational models, build AI using their own data, and run workloads closer to end users. This is critical for real-time AI applications like interactive chatbots or personalised citizen services. Organisations like Eka Care, Max Life, and Yellow.ai are already leveraging this.
Could you share any examples where generative AI is solving unique key business challenges?
Let me start with IndiaMART. It is one of the world's largest B2B marketplaces and had a unique challenge — 60 per cent of users came from Tier-2 and Tier 3 cities where English isn’t the native language. They used our generative AI technology to launch hindi.indiamart.com, transforming their catalogue from English to Hindi, with 500,000 listings now available.
The second example is DTDC. They launched an engagement platform called DIVA 2.0 that transforms how customers and suppliers get shipment information using generative AI. They use Amazon Bedrock and its agentic capabilities to enable autonomous transactions.
The third example is Zomato. Restaurant partners upload over 10,000 images daily. Zomato doesn't use AI to generate fake images but uses AWS Inferentia to enhance real ones — retaining originality while improving appeal. After adopting Inferentia 2, they achieved 25 per cent lower latency and 50 per cent better price performance. This showcases the impact of generative AI across different industries.
You mentioned the country's linguistic diversity. How is AWS supporting the development and deployment of large language models (LLM) which are tailored to the India market?
India is a voice-first country and language plays a very important role. We have AI solutions that understand the language of India. Amazon Transcribe supports more than 10 Indian languages, and we are constantly adding more. That acts as a foundation layer for generative AI applications. Amazon Nova has rich support for Indian languages, and we are constantly improving that. The example I mentioned earlier, IndiaMART, launched in Hindi, and they have plans of making it more regional as well.
What is AWS' role in terms of enabling enterprises as well as startups or developers to build open-source innovations, experiment and deploy GenAI?
Fundamental to this is using Amazon Bedrock. We provide access to open-source models — Mistral or Llama — and have done optimisation specifically for Llama to get better inferencing performance. We continue to harness and provide open-source innovations and not lock down customers to a specific model, so they can harness the capabilities that open-source innovation brings.
I gave you examples of Indian enterprises, and the foundation of this is that we must provide choice to customers. Generative AI is going through rapid innovation, so customers need the right tools for the right job to continuously innovate.
We’re also working on agentic AI, which has a transformative impact. We recently launched Nova Premier, a model that excels at reason, and are building agentic capabilities on top of that.
In December 2024, we launched multi-agent collaboration on Amazon Bedrock. Multiple agents can now coordinate to deliver a task; this will profoundly impact how applications are architected.
We continuously work backwards from customer requirements, understanding what they require and innovating on their behalf. Developers and customers get unfettered access to these technologies, regardless of their size and their technical sophistication.
As India pushes for digital sovereignty and innovation through initiatives like Digital India and India AI mission, what role does AWS envision for itself in supporting these national priorities?
These are exciting times for India’s economy. The government is on a trajectory to build a $1 trillion digital economy by 2025 and develop the country into a $5 trillion economy in the next few years. Embracing cloud technology is critical for India to realise these ambitions.
The government has stated that the core objective of the Viksit Bharat vision is to foster inclusive economic participation among all citizens.
AWS continues to play a pivotal role in supporting India’s digital public infrastructure journey, powering population-scale platforms like Co-WIN, DigiLocker, UMANG, and Poshan Tracker that drive societal impact.
How do you ensure ethical and privacy-conscious deployment of generative AI in sectors like healthcare, finance, and other services. How are you tackling responsible AI?
Amazon Bedrock Guardrails helps us develop ethical AI without customers investing time in building those controls. With Guardrails, customers can create simple policies to ensure models don’t return content involving profanity or other unacceptable things. Customers define what’s unacceptable — it's a big capability.
The second is personally identifiable information (PII) data. For regulated customers, it’s critical that models don’t return PII they aren’t eligible to receive.
We also introduced automated reasoning checks, which provide a mathematical way to ensure the model isn’t hallucinating or returning erroneous information. All these capabilities are in Amazon Bedrock in an easy-to-consume format.
With AI and cloud evolving rapidly, how does AWS plan to bridge the skill gap for developers?
Across India, the need for urgent digital skills training remains a key priority for industry and government. AWS is committed to addressing the digital skills gap and has trained over 5.9 million individuals in India with cloud skills since 2017. We launched AI Ready to equip over 2 million people worldwide with AI skills by 2025, free of cost, and we reached that goal by the end of 2024.
We also have an AWS Generative AI Competency Program to accredit partners, ensuring they have the expertise, experience, and credentials to help customers solve generative AI business use cases confidently.
