Ask Radhakrishna Banavalikar what he does, and he won’t start with technology. He’ll talk about people, about what they need to learn, unlearn, and reimagine in a world that’s moving faster than their training manuals.
“In my career, I’ve had to practice serial learning and unlearning,” he said in his Winner’s Circle conversation. That phrase, in a way, captures everything about modern enterprise AI: systems that learn, forget bias, and rebuild knowledge responsibly.
Krish, who leads AI, analytics, and modernization programs for Tata Consultancy Services (TCS), has spent two decades turning data architecture and human ethics into a single conversation. His job isn’t just modernization; it’s mediation, between technology’s potential and an enterprise’s readiness to use it safely.
The Case for Responsible AI Modernization
Across the industry, modernization used to mean “faster, cheaper, scalable.” Now it’s shifting toward “responsible, explainable, sustainable.” McKinsey’s 2025 State of AI report found that 63 percent of global enterprises plan to increase AI budgets this year, but less than 30 percent have formal frameworks for ethical use or bias mitigation.
Krish sees that gap every day. “AI is only as good as the data and intent behind it,” he said. “We can modernize infrastructure, but if the model reflects cultural or linguistic bias, we’ve just automated the same old problem.”
That statement mirrors findings from the World Economic Forum’s 2024 AI Governance Report, which warned that “English-trained LLMs are exporting Western bias into global systems,” a direct risk for companies operating across Asia, Africa, and Latin America.
The Multilingual Blind Spot
Krish’s own research focuses on language bias in large language models, particularly underrepresented Indo-Aryan languages like Hindi, Marathi, and Sanskrit.
“Even the best LLMs tend to perform poorly when trained on multilingual datasets,” he said. “If a model can’t process sentiment in a regional dialect, it won’t deliver equitable outcomes.”
That observation aligns with data from UNESCO’s Digital Diversity Initiative, which found that over 80 percent of digital content worldwide exists in fewer than ten languages.
The implication? Entire populations are algorithmically underrepresented. “When language itself is a barrier to data accuracy,” Krish said, “you’re not just missing nuance, you’re missing humanity.”
How Enterprises Modernize with Meaning
Krish helps enterprises rebuild their data foundations using frameworks like data mesh and medallion architecture, modern approaches that decentralize data ownership while preserving governance and trust. “The data mesh model lets teams own their data as a product,” he explained. “That’s how you get semantic coherence across systems, a single version of truth that still reflects local context.”
For readers outside the architecture world, Gartner defines data mesh as “a decentralized socio-technical approach to managing data where domain-oriented teams own, serve, and consume data as a product.”
The key word there is socio-technical, technology and people in equal measure. Krish’s modernization strategy merges compliance with culture, making AI both scalable and socially aware.
He’s also realistic about bottlenecks. “Governance is the hardest part,” he said. “The challenge isn’t building models, it’s ensuring they’re traceable, explainable, and defensible.”
That mindset reflects a growing corporate trend identified in Deloitte AI Ethics & Trust Survey 2025: over 70 percent of Fortune 500 firms now have internal AI governance councils tasked with evaluating model transparency and risk.
From Cloud Architecture to Human Architecture
Krish isn’t just modernizing infrastructure; he’s modernizing people. Over his career, he’s mentored more than 300 professionals into AI and cloud leadership roles. His philosophy is that modernization begins with mindset, not migration.
“A system can only be as ethical as the people designing it,” he said. “That’s why we train both engineers and ethicists in the same room.”
This human-centric approach lines up with TCS’s corporate philosophy, which views digital transformation as an ecosystem built on “purpose-driven progress.” It’s a reminder that the enterprise of the future will not only compute better…it will think better.
The Future: Language Equity as AI Strategy
Krish believes that multilingual AI is the next modernization frontier. It’s not about novelty, it’s about representation. “The goal,” he said, “is for every user, in every language, to feel equally understood by technology.”
That’s not just a moral ideal; it’s a business imperative. According to IDC, companies that integrate AI translation and multilingual sentiment analysis into their CX workflows report customer satisfaction increases of 23 percent and conversion rate gains up to 15 percent.
As AI systems evolve, Krish argues that language inclusion should sit beside accuracy and speed as a primary design metric. The model that “understands everyone,” he says, will also serve everyone, and that’s what modernization, in the truest sense, should mean.