Projects
Working prototypes and computational instruments built in the open — for studying law, language, and narrative as evidence anyone can examine. Each is a live tool you can use, not a screenshot.
Bangladesh Legislative Corpus
Quantitative analysis of 227 years of Bangladeshi statute law — citation networks, repeal dynamics, topic grouping across 15 bilingual domains, and 48 cross-lingual legal tropes tracked from 1799 to 2026.
Narrative Deconstruction
Does genre self-organise from classical discourse features alone — no LLMs? 400 short stories, unsupervised clustering via UMAP + HDBSCAN on sentence-level, syntactic, affective and information-theoretic features.
Chakma ↔ Bengali Transliterator
Phonemic transliteration between the Chakma script (Ojhapath) and Bengali Eastern Nagari. Full consonant, vowel, conjunct and digit mapping.
These are working research instruments, rebuilt in the open. They reflect our default to open data, code, and methods.