EuroSciPy2/97
The 'Sci' in EuroSciPy stands for Science (surprised?), and we wanted this year's program to reflect that more than ever! We created dedicated tracks for scientific domains like "Life Sciences & Biomedicine" and "Physical Sciences & Engineering" to put research front and center. 🔬
3/97
The AI/ML landscape is booming! 🤖 To better reflect the diverse work happening in this space, we've split our AI track into two distinct categories.
The AI/ML landscape is booming! 🤖 To better reflect the diverse work happening in this space, we've split our AI track into two distinct categories.
4/97
1️⃣ First, we have "#LLMs, #NeuralNetworks and #AI Development." This track is for the #builders—those creating the new architectures, frameworks, and foundational models. 🏗️
1️⃣ First, we have "#LLMs, #NeuralNetworks and #AI Development." This track is for the #builders—those creating the new architectures, frameworks, and foundational models. 🏗️
5/97
2️⃣ Second, there's "Applied AI and LLM Technologies and Use Cases." This is for the practitioners—those using AI to solve real-world scientific problems. 👩🔬 This split allows us to highlight both foundational work and practical applications.
2️⃣ Second, there's "Applied AI and LLM Technologies and Use Cases." This is for the practitioners—those using AI to solve real-world scientific problems. 👩🔬 This split allows us to highlight both foundational work and practical applications.
6/97
Big news for 2025: We've introduced a peer-review process! This is a major step towards increasing the quality, fairness, and diversity of our program. The goal is to let the community's voice guide our selections. ✅
Big news for 2025: We've introduced a peer-review process! This is a major step towards increasing the quality, fairness, and diversity of our program. The goal is to let the community's voice guide our selections. ✅
7/97
To support our #diversity goals, the peer review was anonymized. Reviewers couldn't see submitter names, gender, or other identifying info. This helps mitigate #UnconsciousBias and focuses the review purely on the proposal's content. 🧑🤝🧑
To support our #diversity goals, the peer review was anonymized. Reviewers couldn't see submitter names, gender, or other identifying info. This helps mitigate #UnconsciousBias and focuses the review purely on the proposal's content. 🧑🤝🧑
8/97
Let's talk #diversity #data. This has been a long journey for us. For years, we had an "underrepresented groups" checkbox in the #CfP, but it was rarely used, especially by women. This left us with a major blind spot. 🤔
Let's talk #diversity #data. This has been a long journey for us. For years, we had an "underrepresented groups" checkbox in the #CfP, but it was rarely used, especially by women. This left us with a major blind spot. 🤔
9/97
A few years ago, we tried to improve by introducing speaker tickets. Our ticketing system always had an optional gender question, so this unlocked a way to get some diversity data on our accepted speakers for the first time. 🎟️
A few years ago, we tried to improve by introducing speaker tickets. Our ticketing system always had an optional gender question, so this unlocked a way to get some diversity data on our accepted speakers for the first time. 🎟️
10/97
But this led to data gymnastics. We had diversity info in our tickets data set, but our proposal info was in a separate submissions data set. Trying to merge these two was a major headache. 🤸
But this led to data gymnastics. We had diversity info in our tickets data set, but our proposal info was in a separate submissions data set. Trying to merge these two was a major headache. 🤸
11/97
Ultimately, this approach was still flawed. The data was incomplete, missing all rejected proposals and poster presenters. We still couldn't answer the crucial question: what is the diversity of our entire applicant pool? ❓
Ultimately, this approach was still flawed. The data was incomplete, missing all rejected proposals and poster presenters. We still couldn't answer the crucial question: what is the diversity of our entire applicant pool? ❓
12/97
We then updated the "underrepresented" question to a dropdown with an explicit "my gender" option, but this change also fell short. The question was still largely skipped, leaving our data incomplete and preventing true accountability.
We then updated the "underrepresented" question to a dropdown with an explicit "my gender" option, but this change also fell short. The question was still largely skipped, leaving our data incomplete and preventing true accountability.
13/97
So this year, we added a direct, optional "gender" field to the #CfP submission form itself. For the first time, we can track the #diversity of ALL submissions, including rejections, withdrawals, and cancellations. This is a huge step forward for #transparency. ✨
So this year, we added a direct, optional "gender" field to the #CfP submission form itself. For the first time, we can track the #diversity of ALL submissions, including rejections, withdrawals, and cancellations. This is a huge step forward for #transparency. ✨
14/97
All this process has resulted in a packed program for #EuroSciPy2025, featuring tutorials, talks, keynotes, a maintainer track, posters, and sprints! More on the specifics later, but first, a special part of our program: the Maintainers Track! 🛠️
All this process has resulted in a packed program for #EuroSciPy2025, featuring tutorials, talks, keynotes, a maintainer track, posters, and sprints! More on the specifics later, but first, a special part of our program: the Maintainers Track! 🛠️
15/97
Let's talk about a special part of our program: the Maintainers Track! 🛠️ It's a space for the folks working behind the scenes on the open-source tools we all rely on.
Let's talk about a special part of our program: the Maintainers Track! 🛠️ It's a space for the folks working behind the scenes on the open-source tools we all rely on.
16/97
Who is it for? The maintainers of @numpy @scipy @matplotlib @pandas_dev etc. The developers who, alongside their research, invest time fixing bugs, adding features, and supporting the #Community
Who is it for? The maintainers of @numpy @scipy @matplotlib @pandas_dev etc. The developers who, alongside their research, invest time fixing bugs, adding features, and supporting the #Community
17/97
The Maintainers track is a platform to:
🤝 Share experiences & best practices
💡 Collaborate on common problems
📚 Learn about new tools & techniques
🌐 Network & build a strong community
The Maintainers track is a platform to:
🤝 Share experiences & best practices
💡 Collaborate on common problems
📚 Learn about new tools & techniques
🌐 Network & build a strong community
18/97
The Maintainers track's program is diverse and practical, with open discussion rounds about current challenges and talks from maintainers sharing their successes and experiences.
The Maintainers track's program is diverse and practical, with open discussion rounds about current challenges and talks from maintainers sharing their successes and experiences.
19/97
A key focus is on sustainable software development and creating an inclusive community. How do we ensure these vital scientific tools continue to be developed and available for everyone in the future?
A key focus is on sustainable software development and creating an inclusive community. How do we ensure these vital scientific tools continue to be developed and available for everyone in the future?
20/97
The track brings together Core Maintainers, regular contributors, Research Software Engineers (RSEs), and anyone who wants to learn more about the world of software maintenance.
The track brings together Core Maintainers, regular contributors, Research Software Engineers (RSEs), and anyone who wants to learn more about the world of software maintenance.
21/97
Why is this so important? Scientific research today depends on this open-source software. The Maintainers Track recognizes this huge effort and provides the support needed to continue this vital work. #OpenSource
Why is this so important? Scientific research today depends on this open-source software. The Maintainers Track recognizes this huge effort and provides the support needed to continue this vital work. #OpenSource
22/97
The Maintainer Track is mostly curated with invited maintainers, but we're open to submissions too! These are reviewed directly by the program committee (no peer review). Want to connect with fellow maintainers? Email us at [email protected].
The Maintainer Track is mostly curated with invited maintainers, but we're open to submissions too! These are reviewed directly by the program committee (no peer review). Want to connect with fellow maintainers? Email us at [email protected].
23/97
Speaking about contacting program team, did you know you can influence our #keynote lineup? We're always looking for inspiring speakers! 🗣️
Speaking about contacting program team, did you know you can influence our #keynote lineup? We're always looking for inspiring speakers! 🗣️
24/97
You can nominate someone for a keynote at any time, even before the CfP opens. Just email [email protected] with their name and why they'd be a great fit.
You can nominate someone for a keynote at any time, even before the CfP opens. Just email [email protected] with their name and why they'd be a great fit.
25/97
Contact details and any extra info are a huge help. You can already nominate keynoters for #EuroSciPy2026!
Contact details and any extra info are a huge help. You can already nominate keynoters for #EuroSciPy2026!
26/97
So, to recap, what does the final #EuroSciPy2025 program look like? As usual:
🧑🏫 Tutorials (Beginner & Advanced)
🎤 Main Conference Talks & Keynotes
🛠️ Maintainer Track
🖼️ Poster Session
🏃 Sprints (Free for all!)
27/97
This structure was a learning moment in our peer review! Some reviewers, new to EuroSciPy, were unfamiliar with our dedicated beginner track. As a result, some great beginner tutorials were flagged as "unfit for the audience." by some reviewers.
This structure was a learning moment in our peer review! Some reviewers, new to EuroSciPy, were unfamiliar with our dedicated beginner track. As a result, some great beginner tutorials were flagged as "unfit for the audience." by some reviewers.
28/97
This highlighted a gap in our onboarding for reviewers, and we had to send out updated instructions mid-review to clarify the conference structure. It’s a process, and we're learning and improving as we go!
This highlighted a gap in our onboarding for reviewers, and we had to send out updated instructions mid-review to clarify the conference structure. It’s a process, and we're learning and improving as we go!
29/97
So here's a pro-tip for future submitters: a high-quality, (absolute) beginner tutorial has a great chance of being selected! We are always looking for great content for this track. #CfP #ProTip #EuroSciPy2026
So here's a pro-tip for future submitters: a high-quality, (absolute) beginner tutorial has a great chance of being selected! We are always looking for great content for this track. #CfP #ProTip #EuroSciPy2026
30/97
This brings us back to the nitty-gritty of the review process. Now that you know the program structure, let's look at how we evaluated proposals…
This brings us back to the nitty-gritty of the review process. Now that you know the program structure, let's look at how we evaluated proposals…
31/97
Anonymization isn't perfect. A common challenge is when authors link to their GitHub profiles in the proposal, which can reveal their identity. We're committed to fairness, but this is a tricky problem in #OpenScience
Anonymization isn't perfect. A common challenge is when authors link to their GitHub profiles in the proposal, which can reveal their identity. We're committed to fairness, but this is a tricky problem in #OpenScience
32/97
For next year, we're considering recommending tools like https://anonymous.4open.science/. This service creates anonymized links to Git repositories, which would be a great help.
For next year, we're considering recommending tools like https://anonymous.4open.science/. This service creates anonymized links to Git repositories, which would be a great help.
33/97
A tougher challenge is in-text deanonymization. Some authors include their names directly in the abstract or bio, or link to videos of their previous talks.
A tougher challenge is in-text deanonymization. Some authors include their names directly in the abstract or bio, or link to videos of their previous talks.
34/97
How will we handle this? One idea is to use a new @pretalx feature that allows volunteers to create a "scrubbed," anonymized version of a proposal for reviewers to see.
How will we handle this? One idea is to use a new @pretalx feature that allows volunteers to create a "scrubbed," anonymized version of a proposal for reviewers to see.
35/97
But this would be a monumental task for our volunteers. Manually redacting every submission is a huge amount of work, and we're not sure if it's feasible.
But this would be a monumental task for our volunteers. Manually redacting every submission is a huge amount of work, and we're not sure if it's feasible.
36/97
This is an open problem we're actively thinking about for #EuroSciPy2026. How can we best ensure a fair, blind review? We're open to #ideas from the #Community!
This is an open problem we're actively thinking about for #EuroSciPy2026. How can we best ensure a fair, blind review? We're open to #ideas from the #Community!
37/97
Want to help? Email us at [email protected] if you'd like to join our team and help with organization or volunteering next year. OR EVEN THIS YEAR!
Want to help? Email us at [email protected] if you'd like to join our team and help with organization or volunteering next year. OR EVEN THIS YEAR!
38/97
To appreciate our new peer-review process, you need to know the old way. In the past, our program team of 5–8 people reviewed every single submission individually. 🥵
To appreciate our new peer-review process, you need to know the old way. In the past, our program team of 5–8 people reviewed every single submission individually. 🥵
39/97
Then, we'd meet and review them all again as a group. An incredible amount of work! We knew we had to find a better, more sustainable way.
Then, we'd meet and review them all again as a group. An incredible amount of work! We knew we had to find a better, more sustainable way.
40/97
Our goal with peer review was to lighten this load. We wanted each reviewer to handle at most 20–30 proposals. But our initial calculations suggested we'd need an unrealistic number of volunteers. It seemed like the program team was destined to do it all again.
Our goal with peer review was to lighten this load. We wanted each reviewer to handle at most 20–30 proposals. But our initial calculations suggested we'd need an unrealistic number of volunteers. It seemed like the program team was destined to do it all again.
41/97
But then, the #community stepped up! We added a question to the #CfP asking if people wanted to be a reviewer, and we were delighted to see 41 people say yes! 🙌
But then, the #community stepped up! We added a question to the #CfP asking if people wanted to be a reviewer, and we were delighted to see 41 people say yes! 🙌
42/97
We had so many volunteers that we even needed to run a quick pre-selection. Not everyone who signed up was able to participate in the end, so we had to re-assign some reviews, but the load per person was still a massive improvement over previous years.
We had so many volunteers that we even needed to run a quick pre-selection. Not everyone who signed up was able to participate in the end, so we had to re-assign some reviews, but the load per person was still a massive improvement over previous years.
43/97
We learned a lot from this first run and will refine our process next year to better ensure reviewers have prior experience with the #EuroSciPy conference format.
We learned a lot from this first run and will refine our process next year to better ensure reviewers have prior experience with the #EuroSciPy conference format.
44/97
Our review process had 3 stages:
Stage 1: Anonymized peer review.
Stage 2: Program team pre-selects clear winners and converts some talks/tutorials to posters. Reviewers can now see other reviews.
Stage 3: Final decisions and tie-breaking. Reviewer activity is minimal here.
Our review process had 3 stages:
Stage 1: Anonymized peer review.
Stage 2: Program team pre-selects clear winners and converts some talks/tutorials to posters. Reviewers can now see other reviews.
Stage 3: Final decisions and tie-breaking. Reviewer activity is minimal here.
45/97
This brings us to the "Poster as a Fallback" option. Converting talks/tutorials to posters isn't new, but we've worked to make the process more transparent for everyone.
This brings us to the "Poster as a Fallback" option. Converting talks/tutorials to posters isn't new, but we've worked to make the process more transparent for everyone.
46/97
A couple of years back, we added an explicit question to the CfP asking if authors were willing to present a poster. This year, we refined the phrasing to be crystal clear, avoiding any ambiguity.
A couple of years back, we added an explicit question to the CfP asking if authors were willing to present a poster. This year, we refined the phrasing to be crystal clear, avoiding any ambiguity.
47/97
This "fallback" question was hidden from reviewers during Stage 1 but was crucial for the program committee in Stages 2 & 3 to build a fantastic and diverse poster session.
This "fallback" question was hidden from reviewers during Stage 1 but was crucial for the program committee in Stages 2 & 3 to build a fantastic and diverse poster session.
48/97
Why do we do this? Converting high-quality talk/tutorial proposals to posters ensures that great work still has a platform at the conference, even if talk slots are limited.
Why do we do this? Converting high-quality talk/tutorial proposals to posters ensures that great work still has a platform at the conference, even if talk slots are limited.
49/97
Posters provide a unique opportunity for direct, one-on-one interaction with attendees. These in-depth discussions can sometimes be even more engaging and lead to more fruitful collaborations than traditional talks.
Posters provide a unique opportunity for direct, one-on-one interaction with attendees. These in-depth discussions can sometimes be even more engaging and lead to more fruitful collaborations than traditional talks.
50/97
And posters can be highly interactive! Many presenters enhance their posters with QR codes linking to live demos, GitHub repos, or supplementary materials. It's a very dynamic format.
And posters can be highly interactive! Many presenters enhance their posters with QR codes linking to live demos, GitHub repos, or supplementary materials. It's a very dynamic format.
51/97
A huge benefit for poster presenters is the Poster Spotlight Session. This is where you get to pitch your poster to the entire conference audience—something only keynotes and lightning talks get to do, as regular talks run in parallel tracks! ✨
A huge benefit for poster presenters is the Poster Spotlight Session. This is where you get to pitch your poster to the entire conference audience—something only keynotes and lightning talks get to do, as regular talks run in parallel tracks! ✨
52/97
Now, let's dive into what our peer reviewers were looking for. We used a weighted scoring system to guide the process, but it was based on qualitative feedback, not numbers.
Now, let's dive into what our peer reviewers were looking for. We used a weighted scoring system to guide the process, but it was based on qualitative feedback, not numbers.
53/97
Reviewers didn't assign scores directly. They chose from radio-button options with English sentences like "I do not recommend acceptance" or "I believe it's well-written and easy to understand." This was then converted to a weighted score on the backend.
Reviewers didn't assign scores directly. They chose from radio-button options with English sentences like "I do not recommend acceptance" or "I believe it's well-written and easy to understand." This was then converted to a weighted score on the backend.
54/97
✅ Recommendation (Weight: 2.0): This was the most important factor. Overall, is the proposal a good fit for EuroSciPy? Is it interesting, relevant, and valuable to our community?
✅ Recommendation (Weight: 2.0): This was the most important factor. Overall, is the proposal a good fit for EuroSciPy? Is it interesting, relevant, and valuable to our community?
55/97
✅ Clarity (Weight: 1.5): Is the proposal well-written and easy to understand? A clear proposal is often a sign of a clear presentation to come.
✅ Clarity (Weight: 1.5): Is the proposal well-written and easy to understand? A clear proposal is often a sign of a clear presentation to come.
56/97
✅ Audience Fit (Weight: 0.5): Does the proposal match the expected expertise of the EuroSciPy audience for the chosen track? (More about this later 👇🏼)
✅ Audience Fit (Weight: 0.5): Does the proposal match the expected expertise of the EuroSciPy audience for the chosen track? (More about this later 👇🏼)
57/97
✅ Originality (Weight: 1.0): Is the submitter an original author or active contributor to the project they're presenting? We prioritized original work and maintainer submissions this year. (Also, more about this later 👇🏼)
✅ Originality (Weight: 1.0): Is the submitter an original author or active contributor to the project they're presenting? We prioritized original work and maintainer submissions this year. (Also, more about this later 👇🏼)
58/97
To address the 'Audience Fit' confusion (see post 27!), we gave reviewers specific guidelines on the expected Python and domain knowledge for each submission type.
To address the 'Audience Fit' confusion (see post 27!), we gave reviewers specific guidelines on the expected Python and domain knowledge for each submission type.
59/97
🧑🏫 For Tutorials, we have two distinct tracks: Beginner and Advanced. Both assume little-to-no specific domain knowledge, as the focus is on learning a tool or technique. The main difference is the expected #Python expertise.
🧑🏫 For Tutorials, we have two distinct tracks: Beginner and Advanced. Both assume little-to-no specific domain knowledge, as the focus is on learning a tool or technique. The main difference is the expected #Python expertise.
60/97
🎤 For Talks, we aim for a balance. Presenters can assume the audience has some-to-expert #Python knowledge, but should assume little-to-no specific domain knowledge. The goal is to make your work accessible to the whole #SciPy #community!
🎤 For Talks, we aim for a balance. Presenters can assume the audience has some-to-expert #Python knowledge, but should assume little-to-no specific domain knowledge. The goal is to make your work accessible to the whole #SciPy #community!
61/97
🖼️ Posters are perfect for deep, specialized scientific topics. Here, presenters can assume up to expert-level domain knowledge, making it the ideal format for presenting complex research.
🖼️ Posters are perfect for deep, specialized scientific topics. Here, presenters can assume up to expert-level domain knowledge, making it the ideal format for presenting complex research.
62/97
It's also worth noting that proposals submitted directly to the Poster track have priority over talks/tutorials that are later converted due to limited schedule space. So if you know your work is a great fit for a poster, submit it as one from the start!
It's also worth noting that proposals submitted directly to the Poster track have priority over talks/tutorials that are later converted due to limited schedule space. So if you know your work is a great fit for a poster, submit it as one from the start!