Google Summer of Code (GSoC) 2026

Announcements
1

Dear all,

Just like in the last years, in 2026, AiiDA is again participating in the Google Summer of Code (GSoC) program under the NumFocus umbrella.

As you can see in the GSoC timeline excerpt below, we will ourselves only find out whether we receive any student slots for GSoC after the accepted mentoring organizations are announced (19th of Feb). Hence, at this moment, we cannot really provide any guarantee that we will be able to secure a slot.

Still, this thread here is for any discussions leading up to GSOC 2026. If you’re interested in working on AiiDA as a student, use it to say hi and ask any questions you may have. Please also see the NumFOCUS guide for GSOC contributors. We will also provide further updates here.

This year, we have one (quite explorative) project proposal: implementing a natural language interface for AiiDA using multi-agent AI. If the project description reads interesting to you, and you feel that you bring the required skills to the table, please voice your interest here. We would be very excited to work with contributors from whom we can learn ourselves about LLMs, agents, MCP, RAG, and other AI-related topics :slight_smile:

One thing we would like to mention is that we have realized that the number of contributions on GitHub usually spikes in the months leading up to GSoC. We also have the gut feeling that many of the PRs appearing in that time are created without the necessary effort and care required for a large code base such as AiiDA. Frequently, tests, continuous integration (CI), documentation builds, or other pipelines start failing, and we have to walk contributors through fixing those issues.

While we are, of course, thankful about community interest and external contributions, we currently don’t really have the human resources to go through this process, as proper review of PRs takes significant time. This is especially relevant with modern AI tools, in which large changes to the code base can be achieved with just a few prompts, and the actual work comes from reviewing (and fixing) the LLM-generated code, as well as verifying its downstream effects.

Hence, we discourage minor contributions that are not bug fixes or improvements that come from actual usage of AiiDA. In addition, having provided minor contributions to the code base is not a requirement to be considered for GSoC anymore. We have removed the corresponding statement from our past GSoC project pages.

For the application process, we would much prefer familiarity with the codebase, its architecture, and ideally a well-thought-out preliminary architectural design plan for the proposed project.

Thanks,

The AiiDA team

Here is the relevant excerpt from the GSOC 2026 timeline:

January 19 - 18:00 UTC

  • Mentoring organizations can begin submitting applications to Google

February 3 - 18:00 UTC

  • Mentoring organization application deadline

February 3 - 18

  • Google program administrators review organization applications

February 19 - 18:00 UTC

  • List of accepted mentoring organizations published

February 19 - March 15

  • Potential GSoC contributors discuss application ideas with mentoring organizations

March 16 - 18:00 UTC

  • GSoC contributor application period begins

March 31 - 18:00 UTC

  • GSoC contributor application deadline
1 Like
Question about GSoC 2026
Looking to get involved early in the GSoC 2026 AI project
2

Thanks for the Clarification i will go through the codebase

3

I’m very interested in the proposed Natural Language Interface for AiiDA using multi-agent AI. I have hands-on experience with LLMs, RAG, AI agents, and agentic workflows, with recent experience creating a simple terminal-based AI vibe coding agent like using local models.

i have been building rag and agentic ai application to check out my github:Adii0906 (Adii0906) / Repositories · GitHub

If you have some time, I would really appreciate it if you could glance over or even give it a go: GitHub - Adii0906/Local-Agent-CLI: A simple terminal-based local AI agent for general-purpose chat and building small applications with user-controlled file permissions

1 Like
4

Hi,

My name is Khadar, and I am a 4th-year Computer Science undergraduate from India. I previously participated in Google Summer of Code 2024, where I worked on building a Retrieval-Augmented Generation (RAG) framework.

I found the project Natural Language Interface for AiiDA using multi-agent AI very interesting. I am currently going through the codebase and familiarizing myself with the project, and I plan to come up with a detailed approach and proposal soon.

Thank you for the clear and helpful instructions—they made it much easier to get started.

Looking forward to contributing and learning from the community.

Best regards,
Khadar

1 Like
5

Hi AiiDA team,

I’m Aryan Srivastava, a CS student. I’m reaching out to express my strong interest in the Natural language interface for AiiDA using multi-agent AI project.

I’ve been an active contributor to NVIDIA’s RAPIDS (cuDF), where I recently handled two significant technical challenges:

PR #20746: I resolved a regression bug affecting floating-point boolean casting by modifying core Python/C++ interop logic and ensuring NaN propagation matched the Pandas API.

PR #20862: I further expanded this work by [implementing comprehensive unit tests for edge-case casting and optimizing the underlying C++ kernels to ensure memory-safe handling of null values during high-performance data transformations.]

Navigating these large, performance-critical codebases has prepared me for the “Advanced” complexity of the AiiDA multi-agent project. I am currently building a multi-agent system using a FastMCP server to expose complex system states to local models. I am particularly excited about your goal of running specialized agents on consumer hardware; I’m currently optimizing local agent workflows on Apple Silicon (Unified Memory) and would love to help design a secure architecture that exposes AiiDA’s Python API as tools via the Model Context Protocol (MCP).

I look forward to discussing a preliminary architectural design with you all!

Best regards, Aryan Srivastava

6

Hey, good evening! @geiger_j
Any news on GSoC 2026 yet?

8

Hi everyone! :waving_hand:

I’m thrilled to introduce myself to this community. I’m Unnati Kadam, an AI and Machine Learning enthusiast currently studying at Fr. Conceicao Rodrigues College of Engineering. I’ve been following AiiDA’s work and am excited to finally introduce myself!

Linkedin - https://www.linkedin.com/in/unnati-kadam-85a76021b/

A bit about me:

I’m passionate about applying technology to solve real-world problems—something I’ve explored through internships in data analytics and machine learning. I recently worked as a Machine Learning Intern at Version Technology, where I assisted in developing supervised ML projects focusing on classification tasks.

Outside of academics, I love participating in hackathons—some highlights:

  • 2nd runner up at E-Summit EGF Hackathon, IIT Bombay (Gaming Track)
  • Winner at Technovation 2024, VESP
  • 5th rank at Azeotropy Predictioneer ML Hackathon, IIT Bombay

I’m currently working with Python, Java, and Linux, and I enjoy diving into data analysis and NLP problems (recently wrote a blog on sentiment polarity prediction for Amazon reviews!). I’m also part of my college’s Entrepreneurship Cell as Content Head.

Why AiiDA?

I was drawn to AiiDA because it sits at the intersection of two things I care about: scientific computing and reproducibility. The idea of helping researchers manage thousands of simulations while tracking full provenance is powerful—it’s exactly the kind of tool that makes computational science more rigorous and transparent.

I’m eager to learn from all of you and eventually contribute meaningfully to a project that makes research more reproducible.

Looking forward to:

  • Learning from this community
  • Finding good issues to start contributing

Happy to be here! Let me know if there’s anything I should read or set up first.

Cheers,
Unnati Kadam

9

Hi all,

Thank you for voicing your interest in AiiDA and GSoC. We are happy to announce that NumFocus has been selected as one of the mentoring organizations for GSoC 2026 :tada:

Please note, however, that Google does not guarantee slots to any specific group, and allocations are determined based on the available budget. Further, as various open-source projects are participating under the NumFocus umbrella, even with the fact that NumFocus was selected as one of the mentoring organizations, we cannot give you any guarantee that AiiDA will be able to secure a slot.

We are currently still internally discussing how to carry out the evaluation and selection process, but we have noted your interest.

To anybody else reading this post who hasn’t replied to the thread yet, please voice your interest in GSoC 2026 here.

We will provide further updates next week.

The AiiDA team

3 Likes
10

Hi everyone,

I’m a computer science student from India, and I’m interested in participating in GSoC 2026 with AiiDA. I’ve started exploring the AiiDA codebase and contribution guidelines to understand the project structure and workflows.

I’m particularly interested in the Natural Language Interface for AiiDA using multi-agent AI project, as it aligns well with my background in Python and machine learning. I’m currently familiarizing myself with the repository and would like to start contributing with small issues or improvements.

I’d appreciate any guidance on recommended next steps or beginner-friendly issues related to this area.

Looking forward to learning and contributing to the community.
Thanks!

11

Hi AiiDA team and @geiger_j,

I’m Kavin Thakur CS (AI & ML) at Chitkara University, India, Minor in AI from IIT Ropar, India, currently Visiting AI/ML Researcher at National Ilan University, Taiwan. LinkedIn

I’m very interested in the natural language interface project. I’ve been reading through aiida-core, aiida-quantumespresso, plumpy, and aiida-pseudo to understand the internals better. One thing that stood out:

get_builder_from_protocol() seems like the safest entry point for programmatic workflow submission (vs constructing inputs manually), and family.get_pseudos(structure=) raising ValueError on missing elements looks like a common pitfall that an NL tool could catch early. Is this the kind of UX gap the team has in mind for this project, or are you thinking more broadly?

I’ve sent an email with a short motivational letter and some initial architectural thoughts. Would be great to hear the team’s perspective on what’s most valuable to focus on.

Kavin Thakur

12

Hi @GeigerJ2 and the team! I’m Aryan(2nd-year CS, contributor to NVIDIA cuDF). I sent over a preliminary architectural proposal via DM a while back focusing on the FastMCP bridge for the Multi-Agent project. I’d love to hear if the team has any initial thoughts or if there’s a specific area of the aiida-core codebase you’d like me to look at while I refine the final proposal for the March 16th opening.