I am pleased to announce the Summer Geometry Initiative (SGI) 2025, to be held July 7-August 15, 2025!
Here's the website: https://sgi.mit.edu --- applications are due February 17, 2025.
The Summer Geometry Initiative (SGI) is a six-week paid summer research program introducing undergraduate and graduate students to the field of geometry processing. Geometry processing has a long history of breakthrough developments that have guided design of 3D tools for computer vision, additive manufacturing, scientific computing, and other disciplines. Algorithms for geometry processing combine ideas from disciplines including differential geometry, topology, physical simulation, statistics, and optimization.
In the first week of SGI, participants will attend hands-on tutorials introducing the theory and practice of geometry processing. During the remaining weeks, participants will work in teams on research projects led by faculty and research scientists in this discipline, while attending talks and other sessions led by visiting researchers. SGI will be held remotely (online) in 2025, but participants are expected to be engaged full-time. Check out these websites to see projects and reports by our past SGI Fellows:
http://summergeometry.org/sgi2021/
http://summergeometry.org/sgi2022/
http://summergeometry.org/sgi2023/
http://summergeometry.org/sgi2024/
Undergraduates and master's level students with background in computer science, mathematics, or other related disciplines are encouraged to apply. SGI is dedicated to the inclusion of individuals of all backgrounds at all levels of geometry processing research. We strongly encourage women and students from underrepresented and underserved groups to apply, including students who come from low socioeconomic means or were the first in their family to attend college. No prior research experience or coursework in geometry processing is necessary to participate in SGI; students who have excelled in the math, science, and/or computing programs available to them are strongly encouraged to apply. All applicants will be considered equitably regardless of race, ethnicity, or gender. Current PhD students and professionals are not qualified to apply.
We will release a separate call for mentors, TAs, and other volunteers.
SGI is also seeking sponsors! If your organization might be interested in sponsoring SGI, please contact me for more information.
NB: US law prevents MIT and SGI from engaging with persons ordinarily resident in certain comprehensively sanctioned countries and with parties blocked by the US Treasury Department. Pending tax considerations associated to the honoraria, all other international students qualify for participation and are invited to apply.
Please forward widely, and feel free to contact me with questions --
- Justin Solomon
Associate professor, MIT Department of Electrical Engineering and Computer Science
Here's the website: https://sgi.mit.edu --- applications are due February 17, 2025.
The Summer Geometry Initiative (SGI) is a six-week paid summer research program introducing undergraduate and graduate students to the field of geometry processing. Geometry processing has a long history of breakthrough developments that have guided design of 3D tools for computer vision, additive manufacturing, scientific computing, and other disciplines. Algorithms for geometry processing combine ideas from disciplines including differential geometry, topology, physical simulation, statistics, and optimization.
In the first week of SGI, participants will attend hands-on tutorials introducing the theory and practice of geometry processing. During the remaining weeks, participants will work in teams on research projects led by faculty and research scientists in this discipline, while attending talks and other sessions led by visiting researchers. SGI will be held remotely (online) in 2025, but participants are expected to be engaged full-time. Check out these websites to see projects and reports by our past SGI Fellows:
http://summergeometry.org/sgi2021/
http://summergeometry.org/sgi2022/
http://summergeometry.org/sgi2023/
http://summergeometry.org/sgi2024/
Undergraduates and master's level students with background in computer science, mathematics, or other related disciplines are encouraged to apply. SGI is dedicated to the inclusion of individuals of all backgrounds at all levels of geometry processing research. We strongly encourage women and students from underrepresented and underserved groups to apply, including students who come from low socioeconomic means or were the first in their family to attend college. No prior research experience or coursework in geometry processing is necessary to participate in SGI; students who have excelled in the math, science, and/or computing programs available to them are strongly encouraged to apply. All applicants will be considered equitably regardless of race, ethnicity, or gender. Current PhD students and professionals are not qualified to apply.
We will release a separate call for mentors, TAs, and other volunteers.
SGI is also seeking sponsors! If your organization might be interested in sponsoring SGI, please contact me for more information.
NB: US law prevents MIT and SGI from engaging with persons ordinarily resident in certain comprehensively sanctioned countries and with parties blocked by the US Treasury Department. Pending tax considerations associated to the honoraria, all other international students qualify for participation and are invited to apply.
Please forward widely, and feel free to contact me with questions --
- Justin Solomon
Associate professor, MIT Department of Electrical Engineering and Computer Science
We are excited to announce the first FastCode Programming Challenge (FCPC) at PPoPP 2025, organized and sponsored by the Fastcode Community. Fastcode Challenge is a student programming competition, and a subsequent workshop at PPoPP, where outstanding submissions are invited to submit papers and give invited talks.
The competition has two tracks.
Track 1: Software Performance Engineering for Irregular Computation (SPE4IC)
Track 1 (the human track) selects two highly challenging problems for parallel programming for human programmers: single-source shortest paths (SSSP) and breadth-first search (BFS). While both of them are fundamental and easy tasks sequentially, they become much more challenging in the parallel context. Achieving effective parallelism is particularly challenging due to the irregular nature of the computations, such as irregular structures of graphs (e.g., sparse, large-diameter graphs), and irregular distribution of edge weights (e.g., non-uniform weights or large weight ranges). Every year, numerous papers are published at top conferences focusing on improving their performance or related problems, yet achieving significant speedup over sequential algorithms still remains an open challenge. Therefore, there are various open directions to explore. In this track, we are interested in efficient implementations to address these challenges in irregular computation using effective software performance engineering.
Track 2: Large Language Model for Parallel Programming (LLM4PP)
Track 2 (the AI track) focuses on leveraging AI models to write fast parallel code. As Moore’s Law diminishes, software performance engineering (SPE) becomes a key technology for programmers to continue scaling computer system performance. However, SPE, especially parallel programming, has been notoriously difficult for average programmers. Meanwhile, recent advancements in Large Language Models (LLMs) have shown great potential in AI-based code generation/completion. In this contest, we explore the opportunities to leverage LLMs for parallel programming. We ask participants to teach and utilize LLMs to solve parallel programming problems, via prompting and model fine-tuning. We also encourage participants to gather training data samples to enhance code models for parallel programming.
Highlights
Multiple awards and prize money are available! US-based students can apply for a dedicated travel grant to cover travel expenses to attend the workshop at PPoPP in Las Vegas, NV.
The competition is fully supported by speedcode.org, which allows you to debug and test your parallel code online, and provides performance analysis tools. You will get immediate results every time you submit the code online, without requiring access to a multicore server. See instructions here: https://speedcode.org/pages/SPE4IC25.html
All participants will submit their methodology as a short paper, which will be peer-reviewed. Accepted papers will be published as FCPC workshop papers along with PPoPP proceedings.
Award-winning submissions and authors of accepted papers will be invited to present their work at the workshop (happening at PPoPP). More submissions will be invited to present posters.
Important Dates
Nov. 20, 2024: Registration and competition starts
Jan. 24, 2025: Competition ends (AoE)
Feb. 1, 2025: paper submission due (AoE)
Feb. 5, 2025: Participants will receive notifications about competition results
Feb. 15, 2025: paper final notification
Mar. 2, 2025: Workshop
More Information for Participants
For more information, please visit our website at: https://fastcode.org/events/fastcode-challenge/
You can join our discord here https://discord.gg/keSUZY8D for interactive discussions and Q&As.
For track 2 participants: if you need computing resources, please contact fastcode.programming.challenge@gmail.com or Xuhao Chen (cxh@mit.edu )
To register, please use the links below:
[Track 1 Registration]
[Track 2 Registration]
If you have any questions about the problems, please contact
The competition has two tracks.
Track 1: Software Performance Engineering for Irregular Computation (SPE4IC)
Track 1 (the human track) selects two highly challenging problems for parallel programming for human programmers: single-source shortest paths (SSSP) and breadth-first search (BFS). While both of them are fundamental and easy tasks sequentially, they become much more challenging in the parallel context. Achieving effective parallelism is particularly challenging due to the irregular nature of the computations, such as irregular structures of graphs (e.g., sparse, large-diameter graphs), and irregular distribution of edge weights (e.g., non-uniform weights or large weight ranges). Every year, numerous papers are published at top conferences focusing on improving their performance or related problems, yet achieving significant speedup over sequential algorithms still remains an open challenge. Therefore, there are various open directions to explore. In this track, we are interested in efficient implementations to address these challenges in irregular computation using effective software performance engineering.
Track 2: Large Language Model for Parallel Programming (LLM4PP)
Track 2 (the AI track) focuses on leveraging AI models to write fast parallel code. As Moore’s Law diminishes, software performance engineering (SPE) becomes a key technology for programmers to continue scaling computer system performance. However, SPE, especially parallel programming, has been notoriously difficult for average programmers. Meanwhile, recent advancements in Large Language Models (LLMs) have shown great potential in AI-based code generation/completion. In this contest, we explore the opportunities to leverage LLMs for parallel programming. We ask participants to teach and utilize LLMs to solve parallel programming problems, via prompting and model fine-tuning. We also encourage participants to gather training data samples to enhance code models for parallel programming.
Highlights
Multiple awards and prize money are available! US-based students can apply for a dedicated travel grant to cover travel expenses to attend the workshop at PPoPP in Las Vegas, NV.
The competition is fully supported by speedcode.org, which allows you to debug and test your parallel code online, and provides performance analysis tools. You will get immediate results every time you submit the code online, without requiring access to a multicore server. See instructions here: https://speedcode.org/pages/SPE4IC25.html
All participants will submit their methodology as a short paper, which will be peer-reviewed. Accepted papers will be published as FCPC workshop papers along with PPoPP proceedings.
Award-winning submissions and authors of accepted papers will be invited to present their work at the workshop (happening at PPoPP). More submissions will be invited to present posters.
Important Dates
Nov. 20, 2024: Registration and competition starts
Jan. 24, 2025: Competition ends (AoE)
Feb. 1, 2025: paper submission due (AoE)
Feb. 5, 2025: Participants will receive notifications about competition results
Feb. 15, 2025: paper final notification
Mar. 2, 2025: Workshop
More Information for Participants
For more information, please visit our website at: https://fastcode.org/events/fastcode-challenge/
You can join our discord here https://discord.gg/keSUZY8D for interactive discussions and Q&As.
For track 2 participants: if you need computing resources, please contact fastcode.programming.challenge@gmail.com or Xuhao Chen (cxh@mit.edu )
To register, please use the links below:
[Track 1 Registration]
[Track 2 Registration]
If you have any questions about the problems, please contact
fastcode.programming.challenge@gmail.com
For any general questions about the competition, please contact Yihan Sun (yihans@cs.ucr.edu).
For any general questions about the competition, please contact Yihan Sun (yihans@cs.ucr.edu).
Dear students,
in case you are interested, I'm forwarding the call of the hackathon organized by the CINI Data Science Lab: this year the hackathon is on log files of the HDFS filesytem.
Best
Francesco
===========
ITADATAHack 2025 – The National Hackathon for Data Enthusiasts
Dear students,
If you're passionate about Data Science, Artificial Intelligence, or Distributed Systems, we invite you to take part in ITADATAHack 2025, the national hackathon promoted by the CINI Data Science Lab in collaboration with Open Data Playground.
Theme 2025:
“Silent Faults, Loud Signals” – a challenge to detect hidden anomalies in system logs using real tools and authentic data scenarios.
What to expect:
● Realistic HDFS datasets
● Palantir Foundry platform + HPC resources from Turin
● Live leaderboard and continuous mentoring
● A unique experience with visibility and career networking opportunities
● Hybrid final on September 9, 2025: in-person in Turin and online for those unable to attend on-site
Who can participate:
● Bachelor’s and Master’s students enrolled at Italian universities
● Graduates from September 9, 2024 onward
● Teams of 2 or 3 participants
Key dates:
● Registration deadline: August 27
● Online hackathon: August 29 – September 1
● Hybrid final: September 9
Registration link: https://buytickets.at/opendataplayground/1494183
Support & community: Join the official Discord channel → https://discord.gg/nwDAyce8dx
This is a great opportunity to learn by doing, to connect with others, and to make an impact. Get involved, spread the word, and step forward – the “silent faults” are waiting to be found.
Best regards,
Team ITADAhack - Laboratorio CINI di Data Science
in case you are interested, I'm forwarding the call of the hackathon organized by the CINI Data Science Lab: this year the hackathon is on log files of the HDFS filesytem.
Best
Francesco
===========
ITADATAHack 2025 – The National Hackathon for Data Enthusiasts
Dear students,
If you're passionate about Data Science, Artificial Intelligence, or Distributed Systems, we invite you to take part in ITADATAHack 2025, the national hackathon promoted by the CINI Data Science Lab in collaboration with Open Data Playground.
Theme 2025:
“Silent Faults, Loud Signals” – a challenge to detect hidden anomalies in system logs using real tools and authentic data scenarios.
What to expect:
● Realistic HDFS datasets
● Palantir Foundry platform + HPC resources from Turin
● Live leaderboard and continuous mentoring
● A unique experience with visibility and career networking opportunities
● Hybrid final on September 9, 2025: in-person in Turin and online for those unable to attend on-site
Who can participate:
● Bachelor’s and Master’s students enrolled at Italian universities
● Graduates from September 9, 2024 onward
● Teams of 2 or 3 participants
Key dates:
● Registration deadline: August 27
● Online hackathon: August 29 – September 1
● Hybrid final: September 9
Registration link: https://buytickets.at/opendataplayground/1494183
Support & community: Join the official Discord channel → https://discord.gg/nwDAyce8dx
This is a great opportunity to learn by doing, to connect with others, and to make an impact. Get involved, spread the word, and step forward – the “silent faults” are waiting to be found.
Best regards,
Team ITADAhack - Laboratorio CINI di Data Science
Jane Street Virtual Estimathon®!
Tuesday, 25th November @ 6:00pm CET
Location: Virtual (on Zoom)
The Estimathon is a mind-bending mixture of math and trivia. In teams, you'll attempt 13 Fermi problems in 30 minutes, ranging from totally trivial to positively Putnamesque. Be prepared to come up with the best set of confidence intervals! Sample problems include:
Do you know how many computers were connected to the Internet on January 1, 1989?
Or how many YouTube videos have more than 1 billion views?
What about the bandwidth beneath the Atlantic Ocean?
Sign up here by Friday, 21st November, if you're interested in attending. Encourage your classmates to register to join us! Swag will be provided to confirmed attendees.
All attendees MUST have access to Zoom accounts and be willing to use them for this event.
Teams: We cannot guarantee you will be on the same team as any friends you list, but we will do our best. Teams will be created once the event starts. The size of teams depends on the number of attendees. Each participant must complete a separate registration form.
We hope to see you there!
___________
Jane Street is a quantitative trading firm with offices worldwide. We hire smart, humble people who love to solve problems, build systems, and test theories. You’ll learn something new every day in our office—whether it’s connecting with a colleague to share perspectives, or participating in a talk, class, or game night. Our success is driven by our people and we never stop improving.
Want to learn more? Check out the latest happenings at Jane Street.
Tuesday, 25th November @ 6:00pm CET
Location: Virtual (on Zoom)
The Estimathon is a mind-bending mixture of math and trivia. In teams, you'll attempt 13 Fermi problems in 30 minutes, ranging from totally trivial to positively Putnamesque. Be prepared to come up with the best set of confidence intervals! Sample problems include:
Do you know how many computers were connected to the Internet on January 1, 1989?
Or how many YouTube videos have more than 1 billion views?
What about the bandwidth beneath the Atlantic Ocean?
Sign up here by Friday, 21st November, if you're interested in attending. Encourage your classmates to register to join us! Swag will be provided to confirmed attendees.
All attendees MUST have access to Zoom accounts and be willing to use them for this event.
Teams: We cannot guarantee you will be on the same team as any friends you list, but we will do our best. Teams will be created once the event starts. The size of teams depends on the number of attendees. Each participant must complete a separate registration form.
We hope to see you there!
___________
Jane Street is a quantitative trading firm with offices worldwide. We hire smart, humble people who love to solve problems, build systems, and test theories. You’ll learn something new every day in our office—whether it’s connecting with a colleague to share perspectives, or participating in a talk, class, or game night. Our success is driven by our people and we never stop improving.
Want to learn more? Check out the latest happenings at Jane Street.
Google Docs
Jane Street Virtual Estimathon - Tuesday 25th November, @ 5:00pm GMT / 6:00pm CET
Join Jane Street for a virtual Estimathon®!
Running an Estimathon virtually requires some pre-work on our part, so please fill out this form if you are able to join. If anything changes and you are no longer able to attend after you've signed up, email…
Running an Estimathon virtually requires some pre-work on our part, so please fill out this form if you are able to join. If anything changes and you are no longer able to attend after you've signed up, email…