AID2E Technical Meeting

Date: July 02, 2026

Phase I

Derek

• PR #55 (artifact cross-referencing between jobs) has been completed.
• Next priority is completing the CLI.
• Created Issue (Link) for the remaining CLI work and assigned it to Amit and Cynthia.
• Potentially merging dev branch into main
  • Initial thought hold off on putting in main until fully functional
  • Useful having most recent functionality in main for Phase II
  • Should be just migrating configuration models branch
  • Is there a specific Panda specific branch? - No
  • Will wait for dRICH PR then direct into main

Baptiste

• B0 FarForward repository working with 2 branches (slurm and panda)
• Uses real resolution calculation running on Panda
• Should this be merged?
  • (Derek) - keep things parallel so that things that are supported can be copied over - run stress tests
  • Keep full FarForward code functional and together before ready to full support with framework
  • Use positions to test with the AID2E framework - remove polynomials

Cynthia

• Working on CLI phase I
• Pushed dRICH examples and other AID2E updates
  • See PR#30: https://github.com/aid2e/AID2E-framework/pull/30
• Get dRICH optimizer script integrated

Cris

• Are any of the reconstruction algorithms going to have critical dependence on change in the design space?
  • No bias being introduced

Phase II

• Amit (See slides)
• Meifeng Suggestion: Nvidia Nemotron as additional open models working in Codex/Claude layer (see slides)
• Google Vertex - fine tuning from Google
• Google AntiGravity - anyone tried?
• James: Will add

Action Items

• Check Action Items from previous meeting
• New action items
  • Once code assist is semi-working - add users for testing
  • Utilization (GPU contention) - effectiveness