Supervisor(s): Pavel Shchekaturov, Bogdan Jeram

Abstract: 

SPARTA, the real-time computer (RTC) platform of the VLT adaptive optics (AO) systems, and the RTC Toolkit for the ELT are two generations of RTC software that will run side by side at Paranal for many years. Maintaining both separately is costly. This internship takes the first practical steps towards bringing the VLT systems to the ELT software level.

The main goal is to clean up legacy SPARTA soft-real-time (SRTC) code and align its structure and coding style with the RTC Toolkit, with the help of frontier AI coding agents and large language models (LLMs) under the student’s review. In this first step the main SPARTA design and architecture are kept unchanged, the VLT environment is maintained, and the tool chain is moved as close as possible to the ELT Development Environment. Every ported component is checked against the original behaviour with regression tests, and the effort saved by the AI tools is measured.

Beyond the cleanup, the student explores a deeper merge of the two RTC generations: running the full SPARTA SRTC cluster software in the ELT Development Environment through the ELT–VLT gateway; changing the RTC paradigm from concentrator tasks to telemetry republishers; and evaluating a common baseline for the SRTC frameworks of the VLT and the ELT, for example moving from RTI Connext DDS to Fast DDS and etc.

The target is one of the AO systems at Paranal, GALACSI or ERIS (to be confirmed at the start). Deliverables are cleaned-up and ported SPARTA components with functional and numerical tests, an evaluation report with a recommended alignment baseline, and guidelines for AI-assisted porting, including which uses ESO software policy permits. As the scope is broad, the exploration items are prioritised with the supervisor at the start. Required skills: C++, Linux and software engineering; interest in AI-assisted development and real-time systems.