Why
Searching a symptom usually lands somewhere between useless and terrifying. The idea was a checker that reads the actual input and answers in proportion to it — which meant the hard constraint was what the thing is allowed to claim, not how well it generates text.
What
It began at a hackathon with two teammates and did not get finished inside the weekend. I took it back afterwards and completed the prototype on my own: the AI layer was my part from the start, and the frontend and backend became mine too.
How
React runs the intake, Node and Express carry the API, and Llama 3 generates the assessment. Age, gender, and symptoms go into the prompt together, so the response is about a person rather than about a word.
Bound what it is allowed to claim
Every response carries the reminder to see a professional. The app offers a preliminary read, and it says so in the same breath.
Ask for what changes the answer
Age and gender travel with the symptoms, because the same complaint does not mean the same thing across them.
Finish it after the weekend ended
The hackathon build was a concept. Making it real meant owning the parts that were never my assignment.