Kalrish Labs / Independent Research
Ideas worth investigating.
Things worth building.
An independent space for research, experiments, software and writing. Things get built here from scratch, in the open, mostly because understanding how something works turns out to require building it.
Currently Investigating
2026- 01 Search & Information Retrieval Inverted indexes, BM25, ranking
- 02 Distributed Systems Queues, workers, rebuildable projections
- 03 Retrieval-Augmented Answering Cited answers over document corpora
- 04 Healthcare Technology Queueing, ABDM, outbreak prediction
Latest Work
Anvesha — the engine that builds search engines
An open-source, modular search platform. Point it at any collection of documents and it will crawl, process, index, search, rank and answer questions over them, with minimal domain-specific code.
002 Project · HealthcarePranSaathi — intelligent patient queue management for government hospitals
Built after seeing government hospitals dealing with serious queue problems. Implements the ABDM API, uses a custom queueing algorithm, and includes an AI model that predicts virus and flu outbreaks four weeks ahead at district level.
The Archive
IndexThe lab is organised into four archives. Most of them are empty right now — they fill up as the work produces something worth recording, and not before.
Status
Aug 2026A place where one person investigates things properly and writes down what happened.
Kalrish Labs is deliberately undisciplined about discipline. Sometimes the work is a search engine built from scratch, sometimes a healthcare system, sometimes reading until a problem makes sense. The subject changes; the approach does not.