From VLSI to Bhashini: Building India’s Tech Stack

You have recently taken over the position of Director at IIIT Hyderabad. What are your immediate and long-term vision for the institution, especially in the context when India is looking towards technological sovereignty?

This is an opportune time for India to seriously pursue technological sovereignty given the geopolitical environment. India must build its own tech stack from the ground up, covering everything from semiconductors, operating systems, middleware, and device drivers to various network and web technology stacks. We are currently far from this goal because the globalization philosophy of the 1990s led us to cease fundamental developments—a path China wisely avoided. IIIT-H is committed to contributing at all levels. At the hardware level, we are enhancing our Center for VLSI Technology, discussing with local organizations like Micron and AMD to build robust programs in VLSI design and fabrication. As India establishes semiconductor fabs, we need highly trained personnel to develop our own processors and GPUs.

On the software side, following Ministry of Electronics discussions, we recognize the importance of an Indigenous Operating System. We plan to start with an open-source kernel, but the critical requirement is to own, develop, maintain, and commercially enhance it over time. Our formal verification team, currently collaborating with DRDO, is simultaneously building capabilities to formally verify these kernel components, ensuring the substrate is secure. Furthermore, we are focused on Indianizing technology using our Language Technology Research Center (LTRC), which initiated the Bhashini project, providing the necessary expertise for local language technology adoption. Overall, IIIT-H is keen and prepared to be a vital part of India's tech sovereignty journey, leveraging our rich heritage of delivering nationally important projects.

How are you planning to further integrate cutting-edge research in areas like GenAI, Web3.0, directly into the students' curriculum to create future-ready tech leaders?

 Our educational model is distinct: the Institute is structured around 29 research centers, not traditional departments, and our students are deeply involved, with hundreds working as Research Assistants (RAs) and often achieving publication credits early in their studies. Regarding GenAI and Agentic AI, we already offer a 14-week, affordable online course on Agentic AI and a course focused on GenAI prompt engineering. Crucially, students are integrated into research projects at centers like the LTRC and AI Research Center, utilizing these technologies for real-world development. For example, Professor P.K. developed SAREL (S-A-R-A-L), an Agentic AI-based system that converts academic research papers into short, accessible video summaries, which can handle many Indian languages. Students engage through hackathons using the SAREL APIs. We are ideally positioned to continually update education by involving students in these high-level activities.

The institution has recently collaborated with Telangana Cybersecurity Bureau to launch Vyuha Labs. What are the immediate and measurable goals of this facility?

Our collaboration with the Cyber Bureau is centered on practical problem-solving. The immediate goal is to automate the analysis of crime calls received on the 1930 distress line. We are leveraging the LTRC's technology (from the Bhashini project) for automatic speech-to-written language translation of these calls. This dramatically automates the process of analyzing crime events to identify hotspots, common fraudulent numbers, and prevailing modus operandi. Simultaneously, we are mining crime data to generate reports on the most common frauds—particularly investment, lottery, and cryptocurrency fraud—to determine which interventions are effective. For investigation support, we intend to deploy the Cybercrime Navigator framework, a solution developed through a PhD thesis. In the long term, we are exploring leveraging our famous computer vision group for forensics, such as analyzing vast camera feeds to automatically identify and track a suspect's movements through a specific area.

How will the lab ensure that the research conducted within the campus is translated into a scalable and a cost-effective solution that can be utilized by police or government to combat emerging cybercrimes?

Scaling is not the role of a university. If a technology proves successful and other jurisdictions want it, we cannot manage the complex deployment. The scalability and long-term maintenance must be managed by a third party. We utilize our Center for Innovation and Entrepreneurship (CIE), which is an incubation center for deep tech companies, to pass on the technology. The successful solution would be transferred to a startup or a third-party company with experience working with law enforcement for implementation, upkeep, feature addition, and achieving wider, cost-effective deployment.

How is IIIT Hyderabad adopting GenAI tools in today's time? And how do you see these tools like a primary teaching aid or a research enabler, or a fundamental shift that requires a change in how computer science is being taught?

The rise of GenAI is both a worldwide concern and an opportunity. While we do not discourage its use, it must not be used for cheating. Consequently, the teaching methods, projects, and homework must fundamentally change. Asking students to write simple programs (like sorting) is obsolete, as GenAI handles these tasks efficiently. Exercises must now require students to thread together components generated by AI into a unique solution, shifting the focus from routine coding to innovative problem-solving and system integration. The primary adoption of GenAI/Agentic AI in the institute is already happening through involving students as RAs in research labs. We have faculty focused on building research centers specifically dedicated to redefining the education system in the context of GenAI, ensuring the tools enhance productivity without encouraging academic dishonesty. We will be briefing the Parliamentary Standing Committee on Education on these specific concepts soon.

Could you elaborate on IIIT-H industry engagement model? What are these different engagement models, such as collaborative research or sponsored projects, that can be utilized by companies outside to leverage deep tech innovations?

Unlike IITs, IIIT-H does not receive operational funding from the government, making us highly dependent on external research projects from both government bodies and industry. We have strong local connections with companies like Microsoft, Amazon, Micron, and AMD. Industry typically funds research through three channels: funding specific short-term projects (1-2 years), leveraging CSR funding for broader cause-based research, or asking for development services, the last of which is undesirable, as institutes are not meant to be cheap software development outfits. I believe a major shift is needed in the relationship. Industry in India often views institutes as bespoke development partners with very short-term goals, rather than partners for long-term capability development. While specific research centers have been established (like the FC Kohli Center by TCS on our campus), industry should primarily invest in long-term, large-scale Capability Centers (5-10 year horizons). This provides impetus for sustained research in directions critical to the company's future, ensuring that when confronted with complex, future-facing problems, they know where to turn for solutions. This model provides significantly more value to the company in the long run.