LLM Data Annotation, Evaluation & QA (Multilingual)
About this Gig
I provide precise, human-in-the-loop data work that makes LLMs measurably better: • Data annotation & labeling — text, image and audio; SFT/RLHF preference data; classification and categorisation. • Response evaluation & QA — rubric-based scoring, side-by-side comparisons, hallucination and safety checks. • Audio & speech — high-precision transcription (including technical/ATC audio) and Mean Opinion Score (MOS) evaluation for TTS systems. • Multilingual — English, Hindi, Urdu and Marathi. Backed by a Master of Computer Applications (MCA) and 3+ years on large-scale LLM data projects (clients including Google and ElevenLabs). I work fully written/async, follow your guidelines exactly, and deliver clean, well-documented data with rigorous quality control.
Requirements
Your annotation/evaluation guidelines or rubric, a small sample of the data (text/audio/image), the target language(s), the tool or format you want (e.g. Label Studio, spreadsheet, JSON), and your total volume plus deadline.
