University of Wisconsin - Madison, High Energy Physics

Position ID:Univ-High Energy Physics-POSTDOC [#27435]
Position Title: 2024 USCMS S&C R&D Initiative
Position Type:Postdoctoral
Position Location:Illinois, United States [map] sort by distance
Subject Area: Physics / Particle Experiment (part-ex)
Appl Deadline:2024/05/19 11:59PMhelp popup (posted 2024/04/02, listed until 2024/06/19)
Position Description:   URMs  

*** the listing date or deadline for this position has passed and new applications are no longer accepted. ***

Call for proposals: 

The U.S. CMS Software & Computing (S&C) operations program seeks proposals from U.S. CMS institutes for the 2024 U.S. CMS HL-LHC S&C R&D Initiative. Successful applicants will be awarded funds (up to 50% remuneration) for supporting postdoctoral research associates who will work within the HL-LHC R&D area in the U.S. CMS Software and Computing Operations Program. We encourage proposals from PIs who are not currently involved in U.S. CMS Software & Computing tasks. 

Proposals are requested to address areas identified as priorities for HL-LHC R&D within the U.S. CMS S&C program such as utilization of object stores, enhancing data delivery robustness, exploring usage of ServiceX, optimizing end-to-end analysis frameworks, GPU-based algorithm development etc. Examples of specific projects are listed at the bottom of this message [1] and are also available at this link together with short abstracts. Interested PIs are encouraged to contact the U.S. CMS HL-LHC R&D area co-leads Lindsey Gray and David Sperka with questions about topics and the application process. To enable PIs who have not yet worked in the area of software & computing to write the best possible proposal, we are happy to match the PI with an experienced mentor from the software and computing area. 

Proposals should be submitted by institute PIs and consist of a cover letter, and a research statement (5 pages maximum). The research statement should summarize the proposed plan of work to be undertaken by the postdoc over a year-long period, the importance of the project and its impact on HL-LHC Software and Computing. A timeline for all major activities including milestones and deliverables over this one year period should be included. The proposal should also describe the postdoc’s past accomplishments and competency and/or present a plan for mentorship and supervision of a potential new hire (if funds allow). For a postdoc who has already been identified, please submit a biographical sketch (limited to two pages). The cover letter submitted as part of the application should specify the requested start and end dates and support needed (postdoc salary and fringe). In addition to salary support, some limited travel funds, if needed, may be requested in the cover letter. The preferred starting date for the project/award is in Summer/Fall 2024 with the award covering an initial period of one year. During this period, up to 50% remuneration for postdoc support will be provided by the U.S. CMS S&C Operations program with the remainder provided by the home institution. The remuneration will be limited to salary and fringe benefits; university overhead will not be paid. Renewals for a second year are possible but require a new proposal to be submitted; these proposals will go through the same review process as others. The research statement, in this case, must include goals and milestones achieved in the prior award in addition to work planned for the forthcoming year, as well as related milestones and deliverables. Renewals beyond the second year may be possible for exceptional cases. 

The deadline for submitting applications is Sunday, May 19th, 2024. Applications will be handled online at A review of the proposals will be conducted soon after the deadline and the results will be communicated to the applicants shortly thereafter. 

The U.S. CMS Software & Computing Operations program is committed to building a diverse team and welcomes applications from universities that are currently not involved in U.S. CMS Software & Computing tasks. 

[1] Project ideas (also available at this link together with short abstracts): 

  • Object Stores

    • Come up with ways that they could be used in analysis/production and understand scaling properties

    • More production like or analysis facility-like demonstrations of the technology in a real usage scenario

      • Data delivery optimization

      • New workflows enabled by object stores

  • Data Delivery Robustness

    • Should be able to open files at other sites reliably 

    • Understand the network / systems level issues and why accessing data is not always reliable. And implement solutions and fixes to the underlying infrastructure. Understand the USCMS storage infrastructure from the analysis perspective.

    • Explore use of Rucio for managing private user data

  • Exploratory usage of ServiceX in CMS analysis

    • Workflow that starts from MINIAOD + NANOAOD, joins columns, and produces high level analysis data (e.g. final histograms)

    • Also column reduction/skimming.

  • Optimizing End-to-End analysis frameworks in columnar format***

    • Better documentation and full examples of how to migrate from imperative programming to library based analysis design

    • With the latest versions of tools (e.g. coffea) and in an optimal way

    • Analysis community needs research

    • How to apply all POG object corrections/calibrations

    • How to convert the columnar data formats back to a format needed for statistical interpretation

    • Central implementation of various analysis input ingredients

    • Avoid people needing to start from scratch for every analysis

  • Integrating GPU-based reconstruction algorithms together into a more complete reconstruction workflow

    •  Identifying host-device transfer bottlenecks and removing them

  • Porting of more resource intensive algorithms to GPU and demonstrating offloading with SONIC

    • Domain or AI solutions

    • Optimizing SONIC workflows with multiple GPU algorithms

    • host-device transfer optimization with SONIC

  • CI/CD for Large AI models

    • Updating of large models through the production system

  • QoS for Tape

    • Optimizing tape access for different use cases, e.g. ReReco campaign where RAW is needed, ReMINIAOD where AOD is needed.

    • Something similar to the ATLAS “Tape Carousel”

Application Materials Required:
Submit the following items online at this website to complete your application:

Further Info:
email address
Department of Physics
1150 University Ave
Madison, WI 53706