Merge pull request #27 from fallow64/2025-fellows

Add 2025 fellows

Author: CannonLock

images/kashika_mahajan.png

@@ -0,0 +1,27 @@
+name: Kashika Mahajan
+title: Fellow
+institution: Morgridge Institute for Research
+status: Past
+organizations:
+  - chtc
+image: images/kashika_mahajan.png
+
+
+fellowship:
+  name: Research Facilitation Project
+  description: |
+    Researchers often have a hard time quickly figuring out how their
+    computational workloads are doing. The way cluster data is currently
+    shown through the system can be overwhelming and hard to
+    interpret, especially for someone who isn't deeply familiar with
+    how everything works. It's not always obvious whether their jobs
+    are running smoothly, stuck, or underperforming.
+
+    That's what this project set out to fix. We wanted to build tools
+    that make it easier for researchers to understand what's going on
+    with their jobs, without having to dig through complicated logs
+    or system outputs. A big part of the process was also figuring
+    out what information researchers actually care about most, and
+    how we could present that in a way that's simple, useful, and
+    clear.
+  mentor: Andrew Owen and Ian Ross

images/sandhya_nayar.jpg

@@ -0,0 +1,29 @@
+name: Sandhya Nayar
+title: Fellow
+institution: Morgridge Institute for Research
+status: Past
+organizations:
+  - chtc
+image: images/sandhya_nayar.jpg
+
+
+fellowship:
+  name: Smart Aggregation and Indexing with the Archive Librarian
+  description: |
+    HTCondor saves large amounts of useful data about finished jobs
+    that can help administrators, facilitators, and researchers
+    who are trying to utilize historical job data. Currently,
+    condor_history manually parses through flat files to get users
+    their data; this makes it significantly harder to use, especially
+    for older jobs, with no easy way to create useful aggregates.
+    
+    This project focused on the creation of the “Archive Librarian” service,
+    to index job records from the HTCondor history archive and
+    improve the usability of historical job data. Each job's “address”,
+    which is the combination of an offset and file metadata of a
+    historical archive file, is stored in a database. This allows
+    the librarian to directly locate job records and avoid using a
+    sequential scan. The database also stores relevant job attributes
+    that researchers commonly rely on, enabling future support for
+    complicated queries and aggregate statistics.
+  mentor: Cole Bollig and Todd Tannenbaum

images/william_jiang.png

@@ -0,0 +1,28 @@
+name: William Jiang
+title: Fellow
+institution: Morgridge Institute for Research
+status: Past
+organizations:
+  - chtc
+  - pelican
+image: images/william_jiang.png
+
+
+fellowship:
+  name: Integrating Pelican and Globus
+  description: |
+    The Pelican Platform is an open-source software platform designed to create
+    data federations by unifying diverse data repositories (e.g. POSIX
+    filesystems, S3) under a single architecture. It differs from the Globus
+    project, whose main goal is to facilitate peer-to-peer data transfer, and it
+    thus lacks the means to plug into cluster scheduling frameworks like HTCondor.
+
+    This project focused on bridging the gap between Pelican and Globus by
+    integrating Globus transfer capabilities into Pelican. This will allow users
+    to leverage Globus's robust data transfer features while benefiting from
+    Pelican's unified data access model. The integration involved developing a
+    Globus backend for Pelican and enabling seamless data transfers between
+    Pelican-managed repositories and external systems via Globus.
+
+
+  mentor: Emma Turetsky and Justin Hiemstra