Interview on KPIT’s Strategy for Autonomous Driving “Europe is Our Benchmark for Safety”

KPIT CTO – AI Gaurav Kakati and Joint Managing Director, Co-founder and Board Member Sachin Tikekar discuss architectural designs, AI governance, and cyber resilience.

The automotive industry is undergoing a fundamental transformation driven by exponential data growth, the complexity of AI systems, and the need to meet stringent safety and cybersecurity standards. AUTOMOBIL-ELEKTRONIK spoke with Gaurav Kakati, CTO – AI at KPIT, and Sachin Tikekar, Managing Director of KPIT, about their strategic focus on autonomous driving (AD) and advanced driver assistance systems (ADAS) – and why Europe is their key test market.

Mr. Tikekar, the increase in data volume and software complexity requires a fundamental redesign of vehicle architecture. How is KPIT responding to this data explosion, and what is your roadmap for balancing cost and innovation?

Sachin Tikekar: The automotive industry is entering a phase where data is the new raw material. Vehicles generate terabytes of information daily, covering areas like ADAS, infotainment, and connectivity. Traditional distributed architectures can no longer handle this volume without skyrocketing costs.

Our roadmap involves transitioning from a distributed to a domain-based architecture, and eventually to a zonal model supported by a centralized high-performance computing system (HPC). This approach enables preprocessing and data aggregation closer to the source, effectively reducing network congestion. Middleware and abstraction layers harmonize various software stacks to avoid redundancies. We are currently in the centralized architecture phase, which is why we are developing the Automotive Smartcore.

Balancing cost and innovation requires upfront investments in software platforms, but we measure ROI through lifecycle savings, accelerated innovation cycles, and faster monetization of vehicle data.

Mr. Kakati, the EU AI Act classifies mobility applications as high-risk. How do you handle this risk-based framework and ensure transparency and human oversight in AI-driven functions?

Gaurav Kakati: We view the classification of mobility applications as high-risk under the EU AI Act not as a limitation but as an opportunity to build trustworthy AI ecosystems. Especially in autonomous driving systems, the safety implications are immense, which is why we pursue a holistic governance approach. This includes continuous lifecycle monitoring of our systems using simulations to validate AI behavior. We also employ explainable AI (XAI) methods that make algorithmic decisions understandable for engineers, regulators, and consumers. Human-in-the-loop controls ensure that ultimate responsibility remains with humans and that critical decisions are not made automatically. Internally, we use “guardrails” to control data at both the input and output ends of AI systems, ensuring all responses and outcomes comply with regulations and maintain transparency, safety, and trust.

Staying with the topic of safety, Mr. Tikekar, Europe sets strict standards with its General Safety Regulation and technical requirements for Level 4 driverless vehicles. How does KPIT ensure its AI-driven ADAS and AD technologies meet these European standards, and how do they differ from those in North America and Asia?

Sachin Tikekar: Europe follows a safety-first philosophy. We ensure compliance with European regulations through three key pillars. First, regulatory intelligence is embedded in our engineering workflows, allowing design decisions to continuously adapt to evolving standards. Second, we use model-based validation frameworks that directly map the behavior of our AI systems to specific safety requirements. Third, we have specialized homologation and cybersecurity teams in Europe to ensure all systems meet high regulatory standards.

 In contrast, North America emphasizes innovation speed, relying on voluntary frameworks and a “fail fast” philosophy, while Asia focuses more on affordability and rapid market entry. KPIT’s neutral position allows us to develop globally adaptable platforms, with Europe’s stringent safety standards serving as our benchmark.

Mr. Kakati, many AI programs remain stuck in proof-of-concept (PoC) cycles. How does KPIT move from PoCs to production-ready AI platforms for next-generation vehicles?

Gaurav Kakati: We’ve long moved beyond the typical PoC cycle. Our platforms are designed to be scalable, reliable, and production-ready. We deploy AI-driven diagnostics and predictive maintenance not just in individual applications but across entire fleets. For validating driver assistance systems, we use large-scale simulation environments that replace billions of physical test kilometers, ensuring high reliability. We also use generative AI tools to create synthetic data and test cases, significantly improving edge-case coverage. Proprietary AI accelerators shorten training and validation cycles. This shift from PoCs to scalable, deployable systems reflects our belief that AI in mobility must be not only innovative but also measurable, certifiable, and road-ready.

Vehicle connectivity increases the attack surface. How do you integrate cybersecurity and functional safety into AI and ADAS development, and what lessons have you learned from industry cyber incidents?

Gaurav Kakati: Cybersecurity cannot be added later – it must be an integral part of development from the start. Our “Security by Design” framework ensures that security considerations are embedded in AI development from the beginning. This includes comprehensive threat modeling to identify and address potential attack vectors early. We use resilient system architectures with redundancy mechanisms to ensure that no single cyber event can compromise vehicle safety or functionality. Additionally, we employ AI-powered anomaly detection to continuously monitor vehicle networks and identify suspicious or malicious behavior in real time. This combination of preventive, robust, and adaptive security measures ensures seamless integration of cybersecurity and functional safety – a critical factor in protecting modern connected vehicles.

Sachin Tikekar: We invest heavily in partnerships with leading cybersecurity firms because expertise alone is no longer sufficient. A key insight is the need for fast OTA update frameworks and multi-layered defense strategies. The challenge is transitioning from rule-based systems to AI-driven systems; if data is poisoned or corrupted, validation results fail. Safety and cybersecurity are co-developed pillars that define the resilience of future mobility.

Finally, Mr. Tikekar, KPIT is growing globally. Talent acquisition and development are crucial. What skills are you looking for to support this global growth, especially given the complexity of AI development?

Sachin Tikekar: Our core philosophy hasn’t changed. We seek people with the right mindset – which today means critical thinking and problem-solving skills. Systems thinking is essential. We want people who are solution-oriented, not just problem identifiers.

 That’s why we invest in AI academies to upskill engineers and work closely with universities to attract young talent early and promote holistic thinking. This strategy ensures KPIT not only hires talent but also nurtures long-term competency ecosystems.

 Gaurav Kakati: A key shift we’re seeing is that problem-solving ability is becoming a critical skill – not just the ability to write code. Coding is increasingly taken for granted, but the ability to design the right systems is crucial. We see younger generations quickly adapting to this level of systems thinking.

 Sachin Tikekar: We want people who are solution-oriented, not just problem identifiers.