GDG RVCE organized DevSprint ’26, a curated series of workshops aimed at helping developers prepare for the Solutions Challenge 2026 through hands-on learning and peer-driven guidance. The series features four key sessions: “Ok Google, Build my AI”, which focuses on practical AI application development and modern tools like RAG; “VibeCoding Done Right”, which teaches effective ways to work with AI tools like Antigravity and avoid common pitfalls; “How to Win a Hackathon”, centered on idea validation and impactful pitching techniques; and “Build Fast, Ship Secure”, which highlights open-source best practices and security considerations such as protecting API keys. Overall, the initiative is designed to equip participants with both technical and presentation skills to build, refine, and showcase their projects effectively. It emphasized practical learning, peer guidance, and real world application of concepts to bridge the gap between theory and implementation.

As part of DevSprint ’26, GDG RVCE organized the “Ok Google, Build my AI” session on 2nd April 2026, a 1.5 to 2 hour hands-on workshop aimed at introducing participants to practical AI application development using a combination of machine learning and large language models. The session began with participants building a cat dog image classifier on Google Colab using transfer learning, giving them exposure to real world model development workflows. This was followed by integrating the classifier into a simple LangChain pipeline, where a Retrieval Augmented Generation (RAG) approach was incorporated using a curated cat dog knowledge context, enabling the system to better interpret and respond alongside image classification. To complete the end to end application, a basic Streamlit dashboard was developed to visualize and interact with the model.

Given the limited duration and the focus on conceptual understanding, participants were provided with partially completed code and guided to fill in key components themselves, ensuring active engagement and hands-on learning rather than passive observation. Throughout the session, students were highly enthusiastic, actively responding to questions, attempting problem solving steps, and engaging with the concepts being taught. Overall, the workshop successfully provided participants with a clear understanding of integrating ML models with LLM frameworks, incorporating RAG techniques, and deploying simple, functional AI applications.

GitHub Repository: https://github.com/Sudarshan2412/SolutionsWorkshop