Research areas

Our research and development work focuses on building reliable, human-centered, and trustworthy AI systems that operate in real-world communication environments.

Human-AI Collaboration

We explore how people work with AI-powered communication systems in practice. Our research focuses on trust, understanding, and effective collaboration between humans and AI, examining how individual differences, cognitive and emotional responses, and system explanations influence user confidence. These insights guide the design of transparent, predictable, and human-centered AI systems that support smoother workflows and wider adoption.

Woman leaning against a wall outdoors while checking her smartphone.
Man riding public transport and using a smartphone while holding an overhead strap.

Trustworthy AI

Building trust is essential for AI in communication. Our research focuses on fairness, transparency, and accountability in AI-driven systems, reducing bias and unintended behaviour using practical evaluation, fairness, and explainability techniques that work in real-world settings. By balancing performance with ethical responsibility, we enable safer and more reliable communication at scale.

Conversational AI

Conversational AI plays a central role in how people interact with businesses. Our research explores how improved context awareness, personalization, and multimodal interaction – across text, voice, and visual signals – can make conversations more natural and effective. By strengthening contextual reasoning and tailoring responses to individual users, we build conversational systems that feel more relevant, reliable, and human-like.

Woman sitting on a sofa and using a smartphone.
Man sitting on a couch and looking at his smartphone.

AI-powered Communication

We develop AI models that improve the performance and reliability of communication networks in real time. Our research focuses on intelligent routing, adaptive orchestration, and predictive network management to reduce latency, errors, and operational overhead. By anticipating issues and automating recovery, our systems enable more resilient networks, better service quality, and reliable message and call delivery.

Fraud Detection

Fraud and malicious activity threaten trust in digital communication. Our research develops AI-driven methods to detect and prevent fraud across communication channels using anomaly detection, behavioural modelling, and predictive systems. Designed for real-time, large-scale environments, our AI enables faster responses to emerging threats and supports more secure, trusted communication.

Woman standing on a city street in sunlight, looking at her smartphone.
Person sitting on a sofa with headphones around their neck, using a smartphone at home.

Spam Filtering

We build scalable AI systems that detect spam and harmful content across text, images, and voice in global communication networks. Our research emphasizes high accuracy, low latency, and robustness against constantly evolving spam tactics. By addressing topic drift and distribution shifts, our models remain effective in high-traffic production environments.

Voice AI

Voice communication brings unique technical challenges at scale. Our research focuses on improving voice quality and intelligence under real-world network conditions through speech enhancement, noise suppression, compression, and unbiased voice processing. Optimized for low latency and high throughput, our voice AI delivers clearer, fairer, and more reliable voice experiences.

Woman relaxing on a sofa and checking her phone while a dog sits nearby.
Woman standing indoors and using a smartphone.

Special Generative Models

We design generative AI models purpose-built for communication platforms. Our research explores domain-specific architectures that prioritize privacy, efficiency, and operational control, including lightweight and privacy-preserving deployment approaches. Optimized for messaging, automation, and telecom workflows, these models enable secure, compliant, and scalable AI capabilities.