ARIA

ARIA is a research assistant that answers with sources, not vibes. I work on it as an AI Engineer at The Binary Holdings.

I did not start this product. I joined a team that already understood the real problem. Intelligence work falls apart when you cannot see where an answer came from.

My job is to keep the model honest. Retrieve first. Speak second. Leave a trail a person can follow.

Why it matters

Most search still dumps a pile of links. Most chat still sounds sure when it is guessing.

ARIA sits in the middle. You ask a real question. It finds the material that actually supports an answer, then tells you what it used. That sounds simple. It is the whole product.

For a company that lives on research, that is the difference between a clever demo and something you can take into a meeting.

My part

I work on the AI side. How a question is understood. How context is gathered. How a response stays tied to evidence.

I care about the last mile. Citations a human can open. Answers that get quieter when the sources are thin. A conversation that still feels fast.

I am not going to unpack the internals here. That work belongs to the team. The bar I hold is simpler. If we cannot show our work, we should not ship the sentence.

What I want next to my name

If ARIA does its job, research gets faster without getting sloppy. That is the contribution I am here to make.