Masters Thesis Internship: AI in Coding Education

The FOSSEE (Free/Libre and Open Source Software for Education) project at IIT Bombay invites applications from motivated Master’s and Dual Degree students across India for a remote, year-long thesis internship. This is a unique opportunity to contribute to YAKSH, FOSSEE’s open-source automated coding assessment platform, and shape the future of AI-driven pedagogy.

The Challenge: Intelligent Feedback for All

While automated grading provides "Pass/Fail" results, it often fails to explain why code is buggy. Our goal is to develop an AI-driven adaptive learning system that provides beginner-friendly, humane feedback without simply handing out the solution.

Research & Scope of Work

Interns will focus on fine-tuning and optimizing open-source Large Language Models (LLMs) to enhance feedback accuracy for programming education. Key research areas include:

• Model Fine-Tuning: Improving performance of models like DeepSeek, Qwen-Coder, and Llama on syntax error simplification and logical flaw diagnosis.
• Knowledge Distillation & Pruning: Reducing model size (distilling 70B/30B models to smaller, efficient versions) for single-language proficiency without losing pedagogical quality.
• Adaptive Sequencing: Developing algorithms for intelligent problem progression based on real-time student performance.
• Benchmarking: Creating and testing against datasets to measure LLM feedback capabilities.

Candidate Profile

We are looking for students who meet the following criteria:

• Academic Status: Currently enrolled in the final year of a Masters (M.Tech/M.E./M.S.) or Dual Degree program in CS, IT, EE, or related fields.
• Technical Expertise: A strong understanding of LLM Architecture (Transformers, Attention mechanisms) and hands-on experience building RAG (Retrieval-Augmented Generation) pipelines.
• Commitment: Must be available for the entire academic year to satisfy thesis requirements.
• Mindset: Passionate about Open Source, Python, and the intersection of AI and education.

Internship Details

How to Apply

Interested students should prepare a research plan (2 paragraphs) outlining their approach to optimizing open-source models for educational feedback.

Please submit your CV, latest transcripts, and research plan by filling up this form. Shortlisted candidates will undergo a screening task and a brief interview.