Jatin S.

Key Skills

Software

Scale AI

Other

Top Subject Matter

LLM prompt verification and multilingual annotation

Video-to-text action narrative annotation for LLM training and evaluation

Top Data Types

Text

Video

Image

Top Task Types

RLHF

Fine-tuning

Bounding Box

Text Generation

Question Answering

Evaluation/Rating

Prompt + Response Writing (SFT)

Freelancer Overview

Prompt Verifier and Multilingual Annotator. Core strengths include Internal and Proprietary Tooling. Education includes Bachelor of Technology, Ajay Kumar Garg Engineering College (2024) and Senior Secondary (Class XII), Silver Line Prestige School (2020). AI-training focus includes data types such as Text and Video and labeling workflows including RLHF and Fine-tuning.

Labeling Experience

Prompt Verifier and Multilingual Annotator

TextRLHF

Performed data annotation and labeling for AI/ML datasets with an emphasis on accuracy and guideline consistency. Evaluated and verified LLM prompts and responses, refining prompt strategies to improve clarity and intent alignment. Worked on multilingual (Hindi & English) annotation to ensure linguistic accuracy and contextual relevance.• Annotated and labeled AI/ML training examples for evaluation and model guidance.• Conducted prompt validation, refinement, and output quality improvement against requirements.• Automated data handling for missing values, duplicates, and transformations using prompt-based strategies.• Generated and optimized prompts for question answering, summarization, and conversational AI tasks.

2026 - 2026

LLM Trainer (Data Annotator/Video Annotation Specialist)

VideoFine-tuning

Converted robotic task videos into human-readable action narratives to create structured training and evaluation samples for supervised fine-tuning. Evaluated model-relevant outputs using predefined rubrics, scoring task understanding, action sequencing, and outcome accuracy. Identified recurring error patterns and edge cases to improve evaluation datasets and reduce label noise.• Produced 1,000+ structured training/evaluation samples from robotic task videos.• Scored outputs (success/partial success/failure) and generated feedback signals using rubrics.• Documented 300+ error patterns, reasoning gaps, and edge cases for benchmarking.• Maintained 95%+ quality and consistency by following annotation schemas and feedback guidelines.

2025 - 2025

Education

Work History