01
Large Language Models
Reasoning, grounding, and evaluation of LLM-based systems.
Research & exploration
Stay curious. Ask better questions.
A collection of interests and ongoing explorations at the edges of AI, knowledge, and real-world applications. These are investigations, rather than a publication record.
Areas of interest
Questions worth following.
02
Computer Vision
Segmentation and recognition, especially for aerial imagery.
03
GraphRAG and Knowledge Graphs
Combining structured knowledge with retrieval-augmented generation.
04
Agentic AI
Tool-using agents that remain observable and controllable.
05
Low-resource Language AI
NLP for Nepali and other under-represented languages.
06
GeoAI
Machine learning on geospatial and remote-sensing data.
07
AI for Tourism
Recommendation and planning systems for travel.
On the workbench
Currently exploring.
Ongoing
GraphRAG for domain knowledge
Comparing graph-structured retrieval with vector retrieval on domain-specific question answering.
Ongoing
Evaluating agentic workflows
Designing trace-level evaluation for multi-step, tool-using agents.
Exploring
Nepali language resources
Surveying datasets and tokenisation behaviour of LLMs on Nepali text.