AskMyCourse

• 

Inspiration

Ever got mislead by ChatGPT because the explanations it gave weren't specific to the standard in the syllabus? And were instead taylored to the standard in the rest of the world? Yuh me too. Enter AskMyCourse, a chat bot that specifically explains the techniques and methods used in the course, which also helps students ask their lecturer to clarify course-specific methods if there's ever a mismatch or confusion.

What it does

AskMyCourse allows teachers to upload their course materials in PDF format. Students can ask course-specific questions, clarify doubts, and get assistance anytime—right from the materials their instructor provided. Think of it as the course personified 😉.

This would allow the course to be learnt in a stream-lined way, the way it was intended and not disrupted by all the noise from the rest of the world, that could be introduced by students using AI models that rely on a mixed standard.

How we built it

We used React, Typescript, TailwindCSS, Vite and ShadCN for the frontend and Jotai for state management. For the backend, FastAPI and Langchain was used.

Lecturers upload lecture notes and any amount of files, which are converted to a vectorized database and stored in a vector database provided by Pinecone, to facilitate RAG. The user can select a course and ask a question, then a similarity search would be done to extract relevant text corpus, which will be fed into a LLM to generate a course-specific response.

Challenges we ran into

-Training the AI to stay contextually relevant to specific course material without going off-topic.

-Ensuring secure file uploads and linking them only to authorized users.

-Optimizing search and retrieval performance for large PDFs and maintaining quick response times in the chat.

-Building a smooth and intuitive UI that works for both students and teachers.

Accomplishments that we're proud of

-End-to-end flow: From uploading notes to answering questions within minutes.

-Clean, modern UI that’s intuitive and responsive.

-Building a system that not only works technically, but also genuinely helps students learn better and faster.

-Seamless integration of authentication, file handling, and AI inference.

What we learned

-Real-world application of secure token-based API calls and managing role-based access.

-How students and teachers interact differently with course content—and how to build for both.

-Importance of user-friendly design in edtech tools.

• Cosine similarity search for retrieval-augmented generation (RAG).

What's next for AskMyCourse

-Integrating directly into learning management systems (LMS), so that the users can directly access it on the university LMS itself.

-Allowing the lecturer to disable the AI in exam/quiz periods.

-Adding support for video lectures with automatic transcript generation and AI QA.

-Letting students upvote useful answers and flag unclear responses.

-Introducing multilingual support to reach a more global student base.

-Analytics for teachers to understand what students are asking and where they’re struggling.

-A mobile app for quick access on the go.

Built With

• fastapi
• langchain
• react
• shadcn
• tailwind
• typescript