GSoC Proposal Submission – LLM-Powered Java/Kotlin to Dart JNIgen Translator
Harshit Kumar
Respected Daco Harkes & Hossein Yousefi
I hope this email finds you well. My name is Harshit Kumar, and I am excited to submit my proposal for Google Summer of Code (GSoC) 2025. My project, "LLM-Powered Java/Kotlin to Dart JNIgen Translator," aims to leverage large language models (LLMs) to automate and optimize the conversion of Java/Kotlin code into Dart JNIgen-compatible code.
Many Dart developers struggle with integrating native Java/Kotlin APIs due to JNI complexities. This project will explore how effectively an LLM can translate Java/Kotlin snippets into Dart code that works seamlessly with JNIgen. The goal is to refine prompting strategies, incorporate automated feedback loops using the Dart Analyzer, and develop a tool that simplifies JNIgen adoption.
Why This Matters-
Reduces manual effort in writing JNI bindings.
-
Improves the developer experience for Dart FFI integrations.
-
Potential to expand into documentation tools (e.g., browser extensions for automatic Dart snippet generation).
I have a strong background in Dart, FFI, and AI-driven automation, along with experience in system design, networking, and open-source contributions. I have worked on AI-powered solutions, Flutter applications, and large-scale system improvements, making me well-suited for this project. My participation in Google Machine Learning Bootcamp, AI-powered defect detection systems, and university IT infrastructure projects further strengthens my ability to execute this proposal successfully.
I would love the opportunity to contribute to Dart and discuss how I can refine my approach to align with your goals. Please let me know if you have any feedback or suggestions. I look forward to your thoughts!
Best regards,
Harshit Kumar
GitHub Profile | LinkedIn | harshitk...@gmail.com