GSoC 2026 – New contributor introduction & onboarding

[Somsubhra-Nandi]

Hi ML4SCI team,
I’m Somsubhra Nandi, interested in contributing to ML4SCI for GSoC 2026.

I’m currently exploring the DeepLense repository and would love guidance on:
• beginner-friendly issues
• recommended improvements for onboarding contributors

Technical Skills

Languages:

• Python, JavaScript

Machine Learning:

• Data preprocessing, feature engineering
• Model training & evaluation (scikit-learn, PyTorch basics)

Applied AI / LLMs:

• Retrieval-Augmented Generation (RAG)
• Text chunking, embeddings, vector databases
• Prompt engineering & API-based LLM integration

Tools:

• Git, GitHub, Jupyter, Linux

Thanks!

[Jaidhuria]

Hello ML4SCI community! 👋

I’m Jai Dhuria, a second-year B.Tech Computer Science student, excited to start contributing to ML4Sci as a GSoC 2026 aspirant.

My primary background is in Machine Learning and data-driven development, with hands-on experience using:

Python, NumPy, Pandas, Matplotlib, Seaborn, Scikit-learn

Building and evaluating ML models for real-world datasets

Data preprocessing, feature engineering, and model analysis

Alongside ML, I also have experience in full-stack development (MERN stack), which helps me work effectively with dashboards, APIs, visualization tools, and collaborative platforms—useful for developing and deploying ML-driven systems.

I’ve worked on projects involving data analysis, prediction models, and visualization, and I’m particularly interested in applying ML to scientific problems such as large-scale data analysis and research-focused workflows.

For GSoC 2026, I’m looking forward to:

Understanding ML4Sci’s workflows and ongoing projects

Starting with beginner-friendly issues and improving incrementally

Writing clean, well-documented, and reproducible code

Learning from mentors and contributing meaningfully to the community

I’d really appreciate guidance on onboarding resources, datasets, or good first issues where I can start contributing.

Excited to learn and collaborate with you all! 😊

— Jai Dhuria

[unnatikdm]

Subject: Interest in Contributing to ML4SCI for GSoC 2026

Hi ML4SCI team,

I’m Unnati Kadam, an AI & ML enthusiast with hands-on experience in building end-to-end machine learning pipelines, including preprocessing, model tuning, and performance optimization. I have a strong track record in national-level hackathons, having been a finalist and winner at events like IIT Bombay’s E-Gaming Hackathon (Predictioneer), IIT Ropar’s Predictathon, and VJTI’s Innovize. I’m passionate about translating real-world problems into scalable AI-driven solutions and have actively contributed to open-source projects and technical communities.

I’m currently exploring the DeepLense repository and would love to contribute to ML4SCI for GSoC 2026. As a beginner to the project, I’d appreciate guidance on:

• Beginner-friendly issues to start with
• Recommended improvements to streamline onboarding for new contributors

Here’s a brief overview of my technical background:

Languages:
Python, Java, C++, SQL, JavaScript

Machine Learning:
Scikit-learn, CatBoost, XGBoost, Pandas, NumPy, Matplotlib, Seaborn
Feature engineering, hyperparameter tuning, model evaluation

Applied AI / LLMs:
Familiarity with Retrieval-Augmented Generation (RAG), text chunking, embeddings, and vector databases
Experience with prompt engineering and API-based LLM integration

Tools & Platforms:
Git, GitHub, Docker, GitHub Actions, Streamlit, Flask, Azure DevOps, Jupyter, Linux

I’m excited to learn, collaborate, and contribute meaningfully to the ML4SCI community. Looking forward to your guidance and suggestions!

Best regards,
Unnati Kadam
unnatikadam50a@gmail.com

[tejaswinidaundkar3-max]

Hello ML4SCI Community 👋

I’m Tejaswini Ramdas Daundkar, a beginner in open source and a prospective GSoC 2026 contributor. I’m currently getting familiar with the DeepLense project and ML4SCI’s work.

I’d really appreciate help with:

• Beginner-friendly issues

• Guidance on how new contributors can start effectively

Background:

• Python

• Git, GitHub, Jupyter

• Basic ML knowledge (currently learning and exploring)

I’m highly motivated to learn, improve, and contribute actively. Looking forward to collaborating with you all!

Thanks 🙌
Tejaswini Daundkar

[Vansh0736]

Subject: Introduction: First-Year Student with Full Stack & DL Skills interested in DeepLense

Hello ML4SCI Community 👋,

My name is Vansh Jain, and I am a first-year engineering student with a strong passion for Space Sciences and Deep Learning. I am writing to express my interest in contributing to DeepLense and the wider ML4SCI organization for GSoC 2026.

My Technical Background:

AI/ML: I have experience working with Neural Networks, Computer Vision (CV), and RNNs. I am also proficient in robust classical algorithms like Random Forests, Support Vector Machines (SVMs), and Gradient Boosting, and I am currently exploring advanced architectures like Vision Transformers (ViTs).

Full Stack Web Dev: I am proficient in JavaScript and TypeScript (MERN stack), which I believe can be useful for any web-based visualization or dashboard tasks required for the models.

Why ML4SCI? I have a keen interest in space exploration and astrophysics, and I am highly motivated to apply my coding skills to solve scientific problems in this domain—specifically the work DeepLense is doing with gravitational lensing.

I would love to get started with some beginner-friendly issues or receive guidance on how to best utilize my full-stack and DL skillset for the current roadmap.

Looking forward to collaborating with you all! 🚀

Best regards, Vansh Jain

[LaxmanPrajapat3]

Hi ML4SCI team,
I’m Laxman Prajapat , interested in contributing to ML4SCI for GSoC 2026.

I’m currently exploring the DeepLense repository and would love guidance on:
• beginner-friendly issues
• recommended improvements for onboarding contributors

Technical Skills

Languages:

Python, JavaScript ,Java , DSA
Machine Learning:

Data preprocessing, feature engineering
Model training & evaluation (scikit-learn, PyTorch basics)
Applied AI / LLMs:

Retrieval-Augmented Generation (RAG)
Text chunking, embeddings, vector databases
Prompt engineering & API-based LLM integration
Tools:

Git, GitHub, Jupyter, Linux
Thanks!

[anshul-119]

Hello ML4SCI team ,
I’m Anshul Sharma, a B.Tech Computer Science student and a GSoC 2026 aspirant. I have a solid technical background and a genuine interest in space exploration and astrophysics, which is what drew me to ML4Sci and particularly the DeepLense project.

As a first step, I’m starting with reproducing one of the existing baseline pipelines to familiarize myself with the DeepLense workflow and better understand the data handling and evaluation setup. Through this, I’m hoping to identify any gaps or possible improvements in reproducibility, documentation, or evaluation clarity.

Relevant experience:

• Python-first ML workflows using NumPy, Pandas, scikit-learn, and PyTorch (basics)
• Hands-on experience with data preprocessing, feature engineering, and model evaluation through coursework, hackathons, and personal projects
• Comfortable working with Jupyter notebooks, Git/GitHub, Docker, and experience programming in C/C++ (academic + problem-solving use cases).
• Familiarity with open-source contribution workflows, including working on pull requests and addressing CI feedback

As I get started, I’d appreciate any pointers on:

• which existing DeepLense baseline would be most suitable to reproduce first
• whether focusing on reproducibility or documentation improvements would be most valuable initially

Looking forward to learning from the community and contributing meaningfully.

[shubhika123]

Hi ML4SCI community,

I’m Shubhika Jain, pursuing BTech in AI/ML. I am exploring ML4SCI and the DeepLense repository as a GSoC 2026 aspirant. I’ve started by going through the repository structure and setting up some of the existing notebooks and baselines on a clean environment to understand the current workflow from a new contributor’s perspective.

While doing this, I’ve been paying close attention to areas that can make onboarding smoother for first-time contributors such as dependency management, setup clarity, and notebook reproducibility. My goal initially is to contribute small but concrete improvements that reduce friction for new users and contributors before gradually moving toward more model- and data-centric work.

At the moment, I’m focusing on reproducing existing pipelines end-to-end and identifying any gaps in setup instructions, assumptions about environments, or silent dependency issues that could block new contributors. I’m planning to open a PR around these aspects once I validate the fixes on a fresh setup.

Going forward, I’m particularly interested in understanding:

• which DeepLense baseline or notebook would be most valuable to reproduce first from the project’s perspective

• whether the team currently prioritizes improvements in reproducibility, documentation, or baseline robustness for onboarding contributors

Looking forward to learning from the community and contributing in a way that’s genuinely useful to the project.

[Riddhima8]

Hi, I’m Riddhima, a computer engineering student with a strong interest in applying machine learning to scientific problems.

I’m particularly interested in contributing to ML4SCI for GSoC 2026, as it sits exactly at the intersection of areas I want to grow in — machine learning, research-driven development, and real-world scientific impact.

Over the past year, I’ve worked on ML-centric projects involving deep learning, evaluation pipelines, and backend systems that support ML workflows. Through GSoC, I’m hoping to deepen my understanding of research-oriented ML, contribute meaningfully to an open scientific community, and learn from experienced mentors.

I’m excited to start contributing and would love guidance on good first issues or project directions to get involved early.

[ammar01-me]

Hello ML4SCI Community,

My name is Sayed Ammar, and I am a B.Tech student in Computer Science Engineering. I am highly motivated to contribute to ML4SCI as part of Google Summer of Code (GSoC) 2026, and I am excited about the opportunity to apply machine learning to real-world scientific research problems.

I have a strong foundation in Machine Learning and Deep Learning, along with hands-on experience in data analysis and model development. My technical skill set includes:
Programming & ML Libraries:
Python, NumPy, Pandas, Matplotlib, Seaborn, Scikit-learn, TensorFlow
Machine Learning & Deep Learning:
Building, training, and evaluating ML/DL models
Working with real-world datasets and performance analysis

In addition to ML, I also have experience in web development and related technologies.
I have worked on projects involving data analysis, predictive modeling, and visualization, and I am particularly interested in applying ML and DL techniques to scientific domains, including large-scale data analysis, simulations, and research-oriented pipelines.

For GSoC 2026, my goals with ML4SCI are to Understand ML4SCI’s research workflows and ongoing projects and to start contributing through beginner-friendly issues and grow progressively
i would great for me to contribute with the ml4sci community and mentors and to make meaningful contributions

I would greatly appreciate any guidance regarding onboarding projects and ML4SCI vision

[ironman1947]

Hello ML4SCI team ,

I’m Om Pradip Chougule, a Direct Second Year student in AI & ML and a GSoC 2026 aspirant. I’ve started exploring the DeepLense repository with the intent of contributing early and practically.

Right now, I’m focusing on reproducing an existing DeepLense pipeline from a clean environment to understand the data flow, evaluation setup, and any friction points for new contributors. My initial contributions will be around improving setup clarity, documentation, and reproducibility before moving toward model-centric work.

Background: Python-centric ML workflows, data preprocessing & feature engineering, scikit-learn and PyTorch basics, experience with Dense Neural Networks and structured ML pipelines. Comfortable with Git/GitHub, Jupyter, and Linux.

I’ll be opening a small PR soon once I validate fixes from a fresh setup. Any guidance on which baseline or notebook is most valuable to reproduce first would be appreciated.

— Om

[Bhavesh007Sharma]

Hi ML4SCI / DeepLense Team,

I am Bhavesh Sharma, a Master’s student in Data Science at Rutgers University with over 5 years of experience in building scalable AI solutions. I am writing to express my strong interest in contributing to DeepLense for GSoC 2026.

I have been exploring the DeepLense repository and am fascinated by the application of Deep Learning to dark matter substructure detection. Given my background in both Computer Vision (published research in CNNs/Image Regression) and Generative AI (LLM Architect), I believe I can make meaningful contributions to the core pipeline and potential multimodal expansions.

My Technical Background & Fit:

Deep Learning & Computer Vision (Core Fit):

Experience designing and customizing CNN architectures (e.g., Inception V3, YOLO).

Published researcher: Authored papers on Image Regression for Autonomous Vehicles and COVID-19 Detection from X-Rays (99.36% accuracy).

Proficient in PyTorch, TensorFlow, and handling large-scale image datasets.

LLMs & Applied AI (Specialization):

Currently an LLM Architect at Phraze, building RAG pipelines using Llama 3, LangChain, and Vector Databases (ChromaDB/Pinecone).

Experience with Fine-tuning (LoRA) and deploying models on AWS/GCP.

Interested in exploring how Large Models can be applied to scientific workflows and data interpretation within DeepLense.

Engineering & MLOps:

Certified AWS AI Practitioner & Oracle Cloud Data Science Professional.

Strong focus on reproducible code, CI/CD pipelines, and API deployment.

I would love guidance on:

Good First Issues: Any specific modules (simulation generation or model optimization) that need immediate attention?

Roadmap alignment: Are there upcoming plans to integrate Transformers or Multimodal models into the lensing pipeline where my LLM background would be valuable?

I am eager to start collaborating and learning from the community.

mail -bhavesh007sharma@gmail.com

Best regards, Bhavesh Sharma

[shlok0603]

Hello ML4SCI Team,

My name is Shlok Nandwana, and I am a Machine Learning enthusiast eager to contribute to the DeepLense project for GSoC 2026.

I have a strong foundation in Python and deep learning frameworks, and I am currently an active contributor to scikit-learn, where I have been working on improving the robustness of the RANSACRegressor by handling non-finite scores in the linear model module.

I am particularly interested in the intersection of deep learning and physics, and I am keen to help improve the codebase and documentation for DeepLense. I’ve already noticed a few open setup issues (like #83 and #85) and would love to help resolve those as I get started with the community.

I look forward to learning more about the project goals and contributing to the ML4SCI community!

[RudyMontoo]

Hi everyone,

I am Rudra Sharma, a 2nd-year B.Tech Computer Science student. I’ve been following ML4SCI’s work for some time and am very excited to formally introduce myself as a new contributor to the organization.

Currently, I am deeply involved in an independent project using the DeepLense dataset to classify gravitational lensing profiles, specifically focusing on the no, sphere (CDM), and vort (Axion) classes.

My current activities include:

Architectural Experimentation: I am training a ResNet-18 model, utilizing its skip connections to detect the very subtle perturbations caused by dark matter subhalos.

Hyperparameter Optimization: I recently completed a manual parameter sweep to identify optimal learning rates and am currently implementing a OneCycleLR scheduler to improve convergence on these physics-based benchmarks.

Physics-Informed Data Pipelines: I am working with .npy data files, ensuring per-image standardization and applying rotational symmetries to maintain the physical integrity of the lensing rings.

While my current focus is on the DeepLense benchmarks, I am highly interested in the broader mission of ML4SCI to apply machine learning to fundamental science. I am eager to start contributing to the community, helping with open issues, and eventually drafting a proposal for GSoC 2026.

I have joined the community channels and am starting with the onboarding documentation now. I look forward to collaborating with you all!

Skills: AI & Data Science (ML, Deep Learning, NLP), Python (PyTorch), Java (DSA), SQL

Best regards, Rudra Sharma

[UzzyDizzy]

Hello ML4SCI Team,

I am Udit Senapaty, 3rd year B.Tech student in Computer Science and Engineering with prior ML research experience, particularly in LLMs and related areas, eager to contribute to DeepLense for GSoC 2026.

I am drawn to DeepLense due to its significant applications in advanced ML techniques such as diffusion models, transformers, and self-supervised learning to strong gravitational lensing, enabling data-driven studies of dark matter and cosmology.

My goal is to contribute by strengthening reproducible model pipelines, improving data preprocessing and benchmarking, and standardizing training and evaluation workflows across DeepLense subprojects. Leveraging my research background in ML, I plan to deliver modular, well-documented code supporting scalable experimentation and reliable scientific inference on realistic astronomical datasets.

An initial guidance would be appreciated for initial contributions and a in-depth roadmap for best use of my existing knowledge and techstack with scope for learning and further contributions aligning with ML4SCI vision.

Best regards, Udit Senapaty

[vishal36-pop]

Hello everyone,

My name is Vishal Reddy Kondakindi and I am a second year undergraduate student studying B.Tech in Artificial Intelligence and Data Science at the International Institute of Information Technology, Bangalore (IIIT-B). I am looking forward to becoming a new participant in Google Summer of Code 2026.

I am most interested in machine learning, deep learning, optimization, and statistical modeling, where I am interested in both theoretical and empirical understanding of models. I like to work on problems where one has to think carefully about data, uncertainty, and generalization, and I am especially interested in open-source projects in the area of scientific machine learning, computer vision, and applied deep learning.

The DeepLense project that seeks to create unsupervised super-resolution techniques of gravitational lensing images to improve image quality and analyze lens properties and sub-structure is of particular interest to me. I was initially introduced to this project in the last GSoC cycle, and its goals are very much aligned with my interests in computer vision and scientific machine learning, which I am currently learning as part of my coursework.

I have practical experience in Python and C/C++ to implement machine learning processes, and I frequently use PyTorch, scikit-learn, NumPy, pandas, and Matplotlib. Coursework and self-directed projects have provided me with a solid background in statistical machine learning, neural networks and deep learning, probability, and optimization.

A key project of mine was to create a pediatric chest X-ray classification system with PyTorch, in which I trained a ResNet-18 model through transfer learning and obtained high precision and recall on pneumonia detection. The project provided me with a hands-on experience in end-to-end deep learning pipelines, such as data preprocessing, model training, evaluation with confusion matrices, and performance analysis.

I have also solved optimization-based machine learning problems under uncertainty, in which I developed and solved convex models, contrasted robust and probabilistic methods, and trade-offs through model behavior and constraints. I have also discussed the issue of class-imbalanced learning, with the emphasis on the right evaluation measure, i.e. F1-score, instead of accuracy.

I hope that via GSoC, I will become a long-term and regular contributor, gain knowledge of large machine learning codebases, and understand best practices regarding experimentation, testing, and reproducibility in open-source ML projects. I would like to begin with issues that are easy to work with and slowly increase the weight of the contributions with the guidance of the mentor.

I am eager to learn with the community and make a contribution.