Arnav Mehra

I’m an engineering student building hands-on conversational AI and supporting model quality through AI data annotation and micro-task evaluation. I’m detail-oriented, comfortable comparing outputs against clear standards, and focused on improving response accuracy and consistency.

In my ongoing annotation work, I complete structured tasks that require careful reading and judgment, and I actively pursue higher-value categories (including Hindi/regional-language annotation) based on task-quality requirements. I’m also consistent about fact-checking claims before accepting them, and I cross-verify information across multiple sources to keep my work reliable.

Alongside that, I develop and iterate on independent voice-assistant systems (FRIDAY and a JARVIS-style assistant) using speech-to-text, language models, and text-to-speech pipelines. I directly evaluate conversational quality, response accuracy, and failure modes while debugging and refining assistant behavior, and I document technical and career topics through well-organized, research-driven reference materials.