Takhti: Researcher-directed software development with AI and open-source tools

Writing in LaTeX usually involves moving between source and compiled output. Frequent compilation catches errors early, but also interrupts writing, particularly for long manuscripts and presentations. This paper describes how a high-energy physics researcher without professional software-engineering training developed Takhti with substantial assistance from ChatGPT and Codex. Initial versions automated conventional compilation; their limitations motivated an incremental live backend based on TeXpresso. The paper explains the checkpoint-and-replay approach, design decisions, and testing on a structured corpus of 81 arXiv source projects alongside other research and teaching documents. A frozen macOS assessment passed the defined interaction checks throughout that corpus, while later Windows tests retained unresolved failures and latency outliers.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-18
DOI
https://doi.org/10.5281/zenodo.24350041
Primary Topic
Software Engineering Research
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Takhti: Researcher-directed software development with AI and open-source tools

Shahin Iqbal
Zenodo (CERN European Organization for Nuclear Research)
Software Engineering Research
preprint

Takhti: Researcher-directed software development with AI and open-source tools

Shahin Iqbal
preprint en

Abstract

Writing in LaTeX usually involves moving between source and compiled output. Frequent compilation catches errors early, but also interrupts writing, particularly for long manuscripts and presentations. This paper describes how a high-energy physics researcher without professional software-engineering training developed Takhti with substantial assistance from ChatGPT and Codex. Initial versions automated conventional compilation; their limitations motivated an incremental live backend based on TeXpresso. The paper explains the checkpoint-and-replay approach, design decisions, and testing on a structured corpus of 81 arXiv source projects alongside other research and teaching documents. A frozen macOS assessment passed the defined interaction checks throughout that corpus, while later Windows tests retained unresolved failures and latency outliers.

Zenodo (CERN European Organization for Nuclear Research)
Abdus Salam Centre for Physics (PK)
Software Engineering Research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon