Research & Innovation
Where our production work starts.
Most of what we ship began as an applied research question. These are the areas we're actively investigating.
Low-resource language NLP
Bangla and other under-resourced languages lack the tooling and labeled data that English-first NLP takes for granted. We build detection, classification and recognition systems specifically for this gap.
Applied computer vision
OCR and document understanding for scripts and layouts that general-purpose vision models handle poorly, including Bangla compound-character recognition.
Responsible healthcare AI
Decision-support models for screening and risk assessment, designed with explainability and human oversight as requirements, not features.
Cognitive & adaptive learning systems
Modeling learner understanding over time to personalize content sequencing and assessment — the research base for our EdTech work.
From Research to Practice
See it applied in our case studies.
Working on a research problem?
If you're a researcher or institution with a computational or AI problem, we'd like to hear about it.