Patchy mobile networks, dropped calls, unreliable data speeds, and the constant juggling between two SIM cards remain everyday realities for smartphone users in India. Add to that the frequent exposure of OTPs and notifications in public spaces, and the gap between how phones are designed and how they are actually used becomes fairly visible.
Speaking at a media roundtable, Kyungyun Roo, Managing Director at Samsung R&D Institute Noida, said these real-world conditions are increasingly being factored into how Samsung develops features for its Galaxy devices. Rather than treating them as post-launch issues, Roo indicated that such use cases are now feeding into the development process itself.
Network Instability and Dual-SIM Use Are Core Inputs
Roo pointed to call performance in unstable network conditions as a key area of focus. He said Samsung has worked on improving how devices handle dropped audio packets by using AI to expand the context window based on prior transmission patterns, allowing short disruptions to be absorbed instead of immediately affecting call quality.
Dual-SIM behaviour, which remains a common usage pattern in India, has also shaped feature development. Roo said scenarios where one SIM loses connectivity while the other remains active have led to solutions such as automatic data switching. He added that the idea of backup calling originated from Indian usage habits and was later extended more broadly across Samsung's devices.
Noida Is Moving From Localisation to Core Development
Roo positioned the Noida R&D centre as part of Samsung's core development structure rather than a localisation layer. The centre contributes across framework, applications, AI, digital health, and enterprise security (Knox), working alongside global teams, particularly in Suwon, under a shared development model.
He said this collaboration is not linear. Different R&D centres work on modules in parallel, with responsibilities split across framework design, feature development, and integration. Issues identified during development or testing are logged on a global system and resolved collectively, rather than being owned by a single centre.
The Noida team is also involved across the product lifecycle, including testing. Roo noted that the centre runs its own test operations and contributes to feature validation before release. This becomes particularly relevant given its proximity to Samsung's manufacturing facility in Noida, where final-stage hardware and software issues are addressed jointly.
For recent development cycles, the Noida team has contributed to areas such as Privacy Display, Now Nudge, Now Brief, Creative Studio, Call Screening, and Direct Voicemail, indicating that its role extends into both user-facing features and underlying framework layers.
AI Development Is Focused on Practical Constraints, Not Just Capability
A large part of Noida's work now sits around AI, but the focus, as described by Roo, is on making these systems usable within real-world constraints rather than simply expanding capability.
Samsung's approach combines on-device and cloud-based AI, with each serving different roles. On-device AI is used where latency, privacy, and responsiveness are critical, while cloud systems are used for more complex processing. Roo said the challenge lies in balancing the two, particularly as expectations from AI features increase.
The Noida centre has been involved in features such as Now Nudge and Now Brief, which rely on contextual understanding and proactive suggestions. These systems are built on what Samsung refers to as a Personal Data Engine, where user data is processed with permissions and largely on-device, giving users control over what is accessed and used.
Testing has also evolved alongside AI development. Roo said the team uses AI-based persona testing in India, where simulated user behaviours are used to validate how features respond in different scenarios. This allows the company to test edge cases and usage patterns at scale before rollout.
Language and Regional Conditions Add Another Layer of Complexity
India's linguistic diversity has also shaped Samsung's AI work. Roo said the company started with Hindi and later added Gujarati, but building these models required addressing dialect variations, informal language usage, and evolving vocabulary, particularly among younger users.
Running these models on-device adds another layer of complexity, as larger AI models need to be compressed without significantly affecting performance. According to Roo, both Hindi and Gujarati currently run on-device, reflecting progress in adapting AI systems to work within device-level constraints.
Beyond language, regional conditions such as high temperatures and varying usage environments also feed into optimisation work, particularly around performance, battery usage, and thermal management.
Privacy and Everyday Use Cases Are Being Reworked
Roo pointed to OTP visibility and notification privacy as another area where user behaviour is shaping feature design. Instead of relying on external privacy films that block the entire display, Samsung's Privacy Display approach focuses on selectively protecting sensitive areas such as notifications.
He said this reflects a shift towards addressing specific user scenarios rather than applying broad, one-size-fits-all solutions.
While much of the industry conversation around smartphones is currently centred on AI capabilities, the examples discussed here point to a more grounded layer of development.
In practice, features are still being shaped by constraints such as connectivity, usage behaviour, and regional conditions. Roo's comments indicate that these factors are increasingly influencing how systems are designed, not just how they are optimised after release.
