DAIR, Dignep's applied AI research unit
DAIR is Dignep’s applied AI research unit, building AI systems for industry, society and public services. It is led by Dr. Yagya Raj Pandeya, Assistant Professor in Kathmandu University’s Department of Artificial Intelligence.
Applied AI research, from prototype to production
DAIR is the research and development arm of Dignep Group. It works on applied machine learning, multimodal perception, optimisation and trustworthy AI, and takes results from prototype to production.
Its work is organised into six applied research areas:
Generative AI and multilingual foundation models
Low-resource NLP for Nepali and regional languages: translation, instruction tuning and domain adaptation.
Retrieval-augmented generation (RAG) for factual assistants that respond quickly, plus safety tooling for hallucination detection, red-teaming and preference alignment.
Affective computing and human-centred AI
Emotion recognition from audio, video and text that holds up on noisy, real-world data.
Behavioural signal analysis (gaze, micro-expressions, prosody) for well-being, education and customer experience.
Privacy-preserving designs and bias assessment across demographics.
Multimodal learning and fusion
Efficient multimodal architectures, self-supervised pretraining and domain-specific models for medical, aerial and agricultural data.
High-dimensional optimisation and compression for edge deployment.
Smart systems and edge AI
On-device intelligence with quantisation, pruning and privacy-aware inference.
Sensing and IoT stacks for monitoring, traceability and infrastructure health.
Decision intelligence that combines machine learning with rule-based systems.
AI governance, safety and policy
Contextual risk matrices, threat modelling, model cards, system cards and audit trails.
Policy research and tooling for responsible deployment in the public sector.
Applied AI for social impact
Problem-driven AI for healthcare, agriculture, education, disaster response and governance.
Co-design with communities and local partners, so systems are inclusive and practical.
Mission
Most AI systems today are built to predict well on a benchmark. DAIR’s aim is AI that learns and adapts in the setting where it is actually used, and that fits human values and local conditions.
Origin
DAIR was set up inside Dignep Group to connect academic research with real-world application.
It is led by Assistant Professor Yagya Raj Pandeya of Kathmandu University’s Department of Artificial Intelligence, who brings more than a decade of experience in deep learning, multimodal data fusion and affective computing. DAIR brings together researchers, engineers and industry partners in Nepal and beyond.
How we approach responsible AI
AI should work with people, not just for them. Three principles guide the unit’s work:
- Local context matters. Nepal’s social, linguistic and infrastructural diversity is a demanding test bed for models that need to work elsewhere too.
- Ethics is part of the architecture. Governance is built into the system rather than added as a policy afterwards.
- Open science. DAIR commits to open-source principles and to releasing frameworks, datasets and benchmarks for community use.
Why a corporate research unit?
- Research that ships. Academic methods paired with product engineering, so validated research becomes working services sooner.
- International benchmarks. Systems built in Nepal to international standards for accuracy, safety and maintainability.
- The full lifecycle. From data collection and model research to MLOps, compliance and monitoring.
- Business priorities. Research is prioritised by measurable business KPIs.
Leadership
Dr. Yagya Raj Pandeya, PhD
Assistant Professor, Department of Artificial Intelligence, School of Engineering, Kathmandu University. Head of DAIR.
Dr. Pandeya completed his PhD and postdoctoral research in AI at Jeonbuk National University, South Korea. At Kathmandu University he leads the Artificial Intelligence and Smart System Research (AISSR) laboratory, which works on AI for agriculture, healthcare, disaster management and cultural preservation.
He was a key member of the team behind Nepal’s National AI Policy (2081) and a team leader for the National AI Strategy (2082), published by the Government of Nepal. More about his work is on his personal site.
Partnerships and Programs
Industry collaborations
We work with startups, enterprises and public-sector organisations to develop AI systems that fit each sector’s needs and constraints.
- Sponsored research. Define a problem statement and run a research roadmap together, with measurable outcomes.
- Joint development. From research prototype to production systems: APIs, microservices and dashboards.
- Pilots and deployments. Controlled rollouts, monitoring and continuous improvement, with safety and governance built in.
Academic collaborations
- Joint projects and co-supervision. Work with faculty and students across departments on shared research goals.
- Exchanges and residencies. Visiting researcher programmes, sabbatical engagements and unit exchanges.
- Shared resources. Datasets, benchmarks, compute and evaluation harnesses for regional problems.
Working with Dignep on AI
If you have an AI problem worth solving, we can help scope it, build it and run it in production.
We work with early-stage startups and with established companies that need to scale.
Machine learning engineers, data scientists, data engineers and data annotators who join your team, focused only on AI development.
We build your AI system and handle the technical work, so you can focus on running the business.
A small team builds and tests your AI product idea, covering the whole project from data to deployment.
Trained annotators label your datasets accurately, so your models learn from data you can trust.
Join a discovery call
We’ll discuss team structure and approach, success criteria, timescale, budget, and required skill sets to see how we can help.
Agree the solution and team
Within days we finalise the project specification, agree an engagement model, and select and onboard your team.
Start and track progress
Once milestones are agreed, work starts. We track progress, report updates and adjust as your needs change.
DAIR is the research and development arm of Dignep Group. It does applied AI research and turns the results into production AI systems. It is led by Dr. Yagya Raj Pandeya of Kathmandu University and connects academic research with real-world use.
Six areas: generative AI and multilingual models, affective computing, multimodal learning, smart systems and edge AI, AI governance and safety, and applied AI for social impact.
Dr. Yagya Raj Pandeya, PhD, Assistant Professor in Kathmandu University's Department of Artificial Intelligence. He was a key member of the team behind Nepal's National AI Policy (2081) and a team leader for the National AI Strategy (2082).
Through sponsored research, joint development and pilot deployments. DAIR works with startups, enterprises and public-sector organisations on AI systems that fit their sector's needs.
Yes. Low-resource NLP for Nepali and regional languages is one of its research areas, including translation, instruction tuning and domain adaptation for multilingual foundation models.
It treats ethics as part of the system architecture, not an afterthought. Governance is built in from the start, using contextual risk matrices, model cards and audit trails.
Have a project like this?
Book a 30-minute discovery call. You'll get a written proposal within two working days.
