Attendees : Jacqueline Keintzel, Ralph Steinhagen, Lex Avstreikh
Distribution/Absent : Sabrina Appel, Penny Madysa, Omar Kamalou
The meeting served as a monthly update for the IRIS project, with participants sharing news and discussing ongoing activities. Adnan began by mentioning an upcoming event at CERN in November and the need to upload data and planning for measurement campaigns. He also announced a training session on the Feature Store during a hackathon in October and a kickoff for the Twinrise project in mid-January. Ralph provided updates on GSI's procurement process for a framework contract related to control system integration and their search for senior candidates with expertise in C++ signal processing. He also discussed export control restrictions affecting their technology. Jacqueline shared updates on her work, including two new PhD students starting soon and plans for measurement campaigns at SuperKEKB in 2027 after this summer 2026 measurement campaign KARA. Lex discussed his involvement in the upcoming hackathon, offering to conduct a hands-on workshop on the Feature Store. The group also briefly discussed the use of ONNX for model deployment.
The meeting participants discussed upcoming events and project updates. Adnan announced a Feature Store training session planned for October 12-13, which may be hybrid and joint between IRIS, TwinRIS, and ARTIFACT teams. A major event in mid-January was discussed, including a TwinRise project kickoff and potential workshop, though scheduling conflicts were noted with Jacqueline's teaching commitment on January 14th. Ralph provided updates on GSI's tender process for control system integration, including time series databases and digital twins, which is currently in progress with a target completion of 3 months. He also mentioned ongoing recruitment efforts for senior-level positions requiring C++ signal processing expertise, with plans to attend conferences in the US and France to identify candidates.
Ralph explained that their technology presents both opportunities and legal challenges. He outlined that while the public and open-source aspects of the framework can proceed, the modeling and specific topology will be restricted due to legal requirements around engaging with certain countries and laboratories. Ralph noted that this situation affects their ability to recruit candidates from restricted countries and requires documentation for the German national export control body, though he remains hopeful about placing the position starting January or February next year during his planned visit to CERN in November.
Ralph and Adnan discussed deliverable responsibilities and payment structures within their project. Adnan clarified that 20% of payments are withheld until final deliverable acceptance, and explained the verification process where reports are reviewed by their team, project management and the EU office before going to the European Commission for payment approval. They confirmed that while there are some ambiguities about specific responsibilities between different teams, GEOFF related work has a beam time planned at GANIL during the project and will have access to surrogate models to support their work.
Jacqueline provided updates on her work, including a presentation she gave at the section leader meeting about her ongoing AI-related projects at CERN, which are now receiving funding through the IRIS initiative. She discussed plans to hire a new PhD student for accelerator performance optimization starting in December or January, and mentioned another agreed student for polarization and monochromatization work. Jacqueline also reported on recent and planned measurements at Kara and SuperKAK facilities, noting that while there is currently no running accelerator at CERN, she expects the next run at SuperKEK to begin in January with an uncertain end date.
The team discussed plans for an upcoming hackathon in October, with Lex offering to conduct a hands-on workshop and potentially a lecture on feature stores. They explored the possibility of organizing a separate workshop for Work Package 2 participants after the TwinRISE in January, though Ralph's availability was limited due to facility recovery from a fire. The group also discussed technical considerations around ONNX as an exchange format, with Lex advising against creating new abstraction layers and recommending existing frameworks like ONNX for better sustainability. A dedicated WP2 workshop could be organized around M6.