Bio Grind

• 

  The layout of the website

• 

  This is the main learning section where children can gain knowledge while reading to the pdf.

• 

  Here we can upload the pdf and can get the probability of the topic to get the maximum marks question

• 

  In this tab we can check our progress and the learning we have done in the past days.

• 

  In this extra section we get the benefit of getting our doubts cleared in just secs while reading a chapter.

• 

• 

• 

• 

• 

Inspiration

Inspired by the need for personalized Class 10 biology prep, we built an AI-powered study platform using Python.

What it does

This AI-powered platform helps Class 10 biology students master board exams. Students upload personal notes and questions to receive instant Previous Year Question (PYQ) importance ratings. The app tracks daily streaks and progress, automatically generating custom daily tasks and targeted practice questions focused entirely on each student's weak areas.

How we built it

​Using Django for user tracking and streaks, LangChain and Sentence-Transformers to process uploaded notes, and FAISS for matching custom questions against previous year questions (PYQs).

Challenges we ran into

Handling messy student notes was tough, but mastering vector databases taught us how to deliver hyper-targeted daily tasks and weak-area practice.

Accomplishments that we're proud of

​Accomplishments We Are Proud Of: Successfully integrating the Gemini 3.8 Flash model into a seamless Retrieval-Augmented Generation (RAG) pipeline that accurately evaluates student-uploaded biology questions against past board patterns while maintaining a reliable streak-tracking backend.

What we learned

We mastered how to optimize modern LLM embeddings and vector databases to deliver hyper-targeted daily tasks, turning messy student notes into structured, real-time adaptive practice.

What's next for Bio Grind

Next for Bio Grind is introducing an AI diagram grader for crucial biology illustrations, adding voice-based oral revision, and expanding into a full "Science Grind" multi-subject platform with a mobile app. We also plan to roll out peer-to-peer quiz challenges based on shared weak areas and custom mock exams.

Built With

• adaptive-question-generation
• and-progress)-*-apis:-gemini-3.8-flash-api-(for-embeddings
• here-is-the-technical-stack-breakdown-for-bio-grind:-*-languages:-python-*-frameworks:-django-(backend-&-user-management)
• importance
• langchain-(rag-&-text-processing)-*-platforms:-web-platform-(pwa-ready)-*-cloud-services:-google-cloud-platform-(gcp)-/-cloud-hosting-*-databases:-faiss-(vector-database-for-pyq-and-note-embeddings)
• postgresql-/-sqlite-(relational-db-for-user-data
• streaks