Vastegune Huiyaan

Inspiration

Undergraduate students especially those in rigorous programs like the BS in Management & Technology (BSMT) face severe academic fragmentation. Students spend hours juggling disconnected spreadsheets for attendance, searching scattered job boards, using generic AI tools that don't understand specific course structures, and managing exam prep across isolated sites.

We set out to eliminate this friction by creating Vastegune Huiyaan: a unified, cohort-tailored operating system that brings coursework, exam prep, career discovery, and community into one seamless dashboard.

What it does

Vastegune Huiyaan is an all-in-one study assistant, semester workspace, and career launchpad designed specifically for undergraduate students:

Vastegune AI: A course-aware study assistant trained directly on BSMT syllabus, case studies, and quiz formats. It handles file uploads (PDFs, Word documents, problem photos) and sits directly inside subjects and mock tests. Semester Workspace: A centralized weekly tracker for lectures, attendance, links, and to-do lists, eliminating the need to rebuild tracking spreadsheets every semester. 400+ Curated Internships: A daily-refreshed database of verified undergraduate internships (including top tech and finance firms like Google, Amazon, Paytm, and Barclays) with direct application links. Global Opportunities Engine: A curated repository of funded summer/winter programs, exchanges, research grants, and global leadership conferences (e.g., HPAIR). Mock Exams & ATS Resume Review: Realistic timed test environments with full performance reviews, alongside an automated ATS resume scanner. Research & Community Hub: A structured paper-writing course with an integrated writing environment, paired with Sus the cohort’s anonymous discussion feed.

How we built it

Frontend & UX: Built a clean, dark-themed responsive workspace application using Typescript, featuring micro-tier subscription logic (Free, Plus, Pro).
Course-Aware AI Backend: Implemented a Retrieval-Augmented Generation (RAG) architecture trained on course materials, slides, and historical quiz structures so the AI understands exact professor expectations and grading schemes.
Document & Vision Processing: Integrated optical character recognition (OCR) and document parsing engines to analyze handwritten problem sets, lecture slides, and PDFs.
Data Pipelines: Engineered automated daily scrapers to verify, filter, and surface direct application links for 400+ undergraduate internships and international funding programs.

Challenges we ran into

Ensuring Precise RAG Context: Generic LLMs often give overly broad answers. We had to fine-tune vector embeddings and retrieval boundaries so Vastegune AI stays strictly aligned with specific BSMT course guidelines and quiz formats.
Maintaining Data Freshness: Job boards change rapidly. Building robust scrapers that verify link validity and filter exclusively for undergraduate-eligible roles required continuous error-handling and validation logic.
Optimizing Sustainable Pricing: Balancing server, database, and LLM API costed us nothing and we kept the platform accessible to students required careful freemium tier engineering (from ₹0 to ₹319/month).

Accomplishments that we're proud of

*We launched this in IIT Jodhpur live in front of whole batch in the official live seminar. Got recognition from our program manager, positive response from our professors and program director of IIT Jodhpur.
High-Accuracy Course Intelligence: Training an AI assistant that can parse complex management and tech cases and deliver tailored explanations that match class formatting. Also, we all created this without spending any extra penny other than domain charge.
UNHEARD Pitch Challenge Integration: Hosting the Breaking Stereotypes! online workshop and pitch challenge for teenagers, providing paid internship opportunities directly through the platform.

What we learned

Cohort Specificity Wins: Generic productivity and AI tools lack context. Tailoring the AI model to exact degree paths and university structures dramatically increases student adoption and trust.
Frictionless Discovery is Essential: Students lose hours looking for valid internship links. Providing clean, direct links to official employer pages delivers immediate, high-value impact.

What's next for Vastegune Huiyaan

Multi-Cohort & Multi-University Expansion: Scaling the underlying course-ingestion pipeline to support additional specialized degree programs and university cohorts.
Real-Time Peer Collaboration: Launching interactive group study rooms, shared lecture notes, and collaborative research drafting directly within the workspace.
Employer Partner Connect: Partnering directly with hiring teams to match top-performing students from the platform with exclusive, fast-tracked internship referrals.

AI usage

We used AI coding tools, mainly Claude (Claude Code), to write most of the website's code, including the frontend interface, page layouts and much of the application logic.

The human team set the product's direction and design: what to build, for whom, and how each feature should work. We set up and configured the infrastructure ourselves, including Vercel hosting, the Supabase database, authentication, domains and email. We also reviewed, tested and deployed every change.

The product's AI features (Vastegune AI and the resume review) also use AI models at runtime. We used gemini api for that.