© Thinh Ngo, 2026
← All work
02

Achievia

AI study-abroad coaching for Vietnamese students — built for the speaking and writing their schools skip.

Event
LotusHacks 2026 (HackHarvard × GenAI Fund)
Result
Top 20 of 200
Duration
36 hours
My role
Pitch lead, market research, live demo
Team
4 — I owned the narrative layer
Stack
ElevenLabs, Bright Data, Manus AI

TL;DR

36-hour build unifying IELTS/SAT prep, university matching, and scholarship discovery. I owned the pitch, market research, user framing, and the live demo. Top 20 of 200.

The user

VN public-school English drills grammar, reading, and listening — rarely speaking or writing. The gap shows at application time: interviews, essays, scholarships. We scoped to those students and built the Speaking Coach (ElevenLabs voice) around them.

What I did

I didn’t write code — I owned the narrative layer: market research, user framing, the pitch, and the live demo. Ran it end-to-end and fielded judge questions solo. Translating technical work for a non-technical room is my core skill.

Hardest 36 hours

Halfway in we were proving the problem, validating users, designing, and building all at once — on no sleep — and decisions started slipping. We took a walk, reset, and made our best scoping calls right after.

Honest framing

No live demo to link — a teammate locked the repo post-event. The proof is the deck, the write-up, and the Top 20 result. Next time I’d treat the hackathon as networking, not just competition.

Artifacts

Next project

WWS