Abhishek Savita

I've contributed to AI evaluation and data workflows at FlyRank AI, Outlier, and OneForma, spanning LLM response assessment, synthetic speech evaluation, Hindi-English transcription, and multimodal data collection.

At FlyRank AI, I design AI-assisted workflows with React, Vite, Claude, and ChatGPT. I evaluate outputs for reasoning, factual accuracy, relevance, instruction following, completeness, and clarity, while documenting reusable prompt templates and evaluation guidelines.

At Outlier and OneForma, I support multilingual AI datasets through structured annotation, TTS and audio-quality review, human-versus-synthetic speech assessment, transcription, and image and video collection.

I also build practical AI and machine learning projects, including an OCR-powered document summary assistant using Python, Flask, and Gemini API, plus a customer churn prediction workflow using Scikit-learn and Power BI.