PhoFilledHackersProject

This stage is where students start to chat with chatbot.
If AI cannot detect, AI will point out that reason
If it's possible, AI provide list universities
Students can compare their chosen universities.
Profile of students is created based on the conversation

Inspiration

Our project is directly inspired by the speech delivered by the leader from ETEST. We realized that students and parents desperately need clearer answers regarding which US universities are currently realistic, how to improve their profiles, and whether their family's financial capacity can meet the requirements. Meanwhile, ETEST counselors are wasting too much time reconstructing context from unstructured chat instead of reviewing a clean, standardized profile to act on for in-depth consultation.

What it does

ETEST Compass is a student-facing admissions advisory web app for Vietnamese students applying to US universities. It performs the following core functions:

Converts raw student input into a structured admissions profile.
Recommends universities across Reach / Target / Safety tiers with transparent reasoning, including budget fit and potential blockers.
Packages the profile and hands off qualified leads to ETEST counselors for paid consultation.

How we built it

We built the system using a modern, scalable stack:

Architecture & Frontend: Turborepo, Next.js 16 (App Router), and shadcn/ui.
Backend & Database: Supabase Postgres and Drizzle ORM, with APIs documented using OpenAPI.
Authentication: Better Auth supporting Google OAuth, Facebook OAuth, and email auth.

Challenges we ran into

We faced and overcame three core challenges:

Understanding the ETEST business problem and ecosystem: The biggest challenge was designing a "Human-in-the-loop" product that supports counselor workflows without replacing human judgment.
Sourcing US university data: Although the MVP scope requires an offline catalog import pipeline, manual data collection is unfeasible. To solve this, we used Codex to generate 5 sub-scripts to automatically scrape and generate data from official admissions, tuition, and scholarship pages.
Balancing AI, UI/UX, and business profitability: We had to apply AI and "transparency by default" design principles—explaining why a school is recommended or downgraded rather than silently gatekeeping—while structuring user flows to capture booking requests, thereby maximizing conversion rates and profits for ETEST.

Accomplishments that we're proud of

Successfully building a seamless pre-consultation workflow that completely resolves the clutter of unstructured student information gathering.
Creating a recommendation engine that clearly distinguishes between "Current Outlook" and "Projected Outlook" based on explicit, user-stated milestones.
Integrating a Chat Assistant that explains recommendation reasoning using structured, source-backed facts rather than relying on unconstrained AI hallucination.

What we learned

We learned that a successful EdTech product relies not just on smart algorithms, but on transparency. Users need to understand exactly why a university is financially or academically out of reach. Furthermore, protecting the business's revenue stream—by effectively handing off qualified leads to counselors—is vital to the project's sustainability.