studentfolio
learn anything, master everything
Created on 22nd June 2025
•
studentfolio
learn anything, master everything
The problem studentfolio solves
Personalized mentorship at scale
People often need instant, contextual feedback whether studying, debugging, or learning something new. Without access to a live tutor or peer helper, progress stalls.
Bridging the “learning gap” in real time
When learning complex material (e.g. code, technical docs, math), confusion often strikes in the moment. Waiting for peer review or instructor replies disrupts momentum and motivation.
Challenges we ran into
1.Reliable Voice Capture & VAD
Mic input varies; distinguishing speech vs background noise and detecting when to start/stop listening is tricky.
Calibration across different hardware and acoustic environments adds complexity.
2.Accurate Transcription & Fallbacks
STT errors lead to misunderstandings; API latency adds delay.
You mitigate this by layering services (Sarvam), but integrating them and handling failures robustly is engineering-heavy.
3. Multimodal Processing with LLM
Combining text (speech transcript) and visual input (real time video streaming) for coherent answers is non-trivial. Synchronizing them into one prompt requires careful design.
Progress made before hackathon
Progress Made Before the Hackathon: Research
Before the hackathon, significant groundwork was laid through comprehensive research to deeply understand the fundamental problems in modern education and the feasibility of using generative AI for real-time, contextual mentorship. Key areas explored include:
Identifying the Core Educational Gap
Analyzed data from global education reports (OECD, UNESCO, etc.) to understand the mismatch between learning outcomes and industry needs.
Studied the limitations of traditional feedback mechanisms in classrooms and online platforms—highlighting the lack of instant support and personalized attention as major bottlenecks to learner progress.
Feasibility of AI-Powered Mentorship
Conducted a technical survey of large language models, speech-to-text, and vision-language systems capable of enabling real-time, multimodal interaction.
Evaluated TTS/STT tools (e.g., Whisper, Sarvam API), visual context tools (screenshot analysis), and prompt engineering strategies for contextual assistance in educational workflows.
User Personas & Learning Pain Points
Interviewed target user groups (students, self-learners, bootcamp participants) to map real-world friction points: delayed help, complex course materials, language barriers, and lack of mentorship.
Validated the need for context-aware AI assistance that integrates voice, visuals, and curriculum-based guidance.
Tracks Applied (2)
Sarvam AI Track
Sarvam.ai
Google Cloud Platform Usage
Google Cloud Platform
Cheer Project
Cheering for a project means supporting a project you like with as little as 0.0025 ETH. Right now, you can Cheer using ETH on Arbitrum, Optimism and Base.