< back_to_work
EwnetCheck · 2024
Misinformation detection platform
A fact-checking platform pairing custom NLP models with a frontend that makes model output readable.
- Role
- Full-Stack Developer
- Year
- 2024
- Stack
- Next.js, FastAPI, TensorFlow
EwnetCheck is a fact-checking platform built around the misinformation models from my low-resource NLP work. The hard problem it solves is legibility: a classifier score means nothing to a non-technical reader, so the interface has to translate model output into something a person can trust and act on.
I built the complete user-facing platform in Next.js, connecting to the model layer through FastAPI. Confidence levels, flagged claims, and supporting signals are presented in plain language with visual weight matched to certainty, so the product never overstates what the model knows.
> key_details
- Full user-facing platform built in Next.js over a FastAPI model layer
- Model confidence translated into plain-language, trust-calibrated UI
- Built on custom misinformation models for Ethiopian languages
Next projectFlyMark →