Purushottam Mani Tripathi

I've trained and evaluated LLMs for Innodata, Scale AI, and SOUL AI, contributing labeled datasets, response rankings, and structured feedback for model improvement.

At Innodata, I annotate text, image, audio, and video datasets; compare model outputs for factuality, safety, relevance, coherence, and helpfulness; and support multilingual Hindi-English transcription. I've also worked on prompt evaluation, rewriting, summarization, fact-checking, and policy-compliance review using Data Compute, SRT, and Multimango.

Previously at Consultadd, I managed IT project workflows, vendor coordination, documentation, schedules, KPIs, and risk tracking. I bring an engineering background alongside practical experience turning evaluation criteria into high-quality AI training data and actionable quality reports.