Sharmilaa saran

Generated prompts and assessed AI-written content for grammar, accuracy, clarity, and relevance to improve model outputs and evaluation metrics.

Transcribed market research interviews in Tamil and English with high accuracy to produce clean datasets for analysis and reporting.

Evaluate and compare model A/B outputs for factuality, usefulness, and instruction compliance to determine better responses and contribute to RLHF leaderboards.

Draw bounding boxes and label sanitary product details in CVAT, check for image issues, and rate image quality to support computer vision model training.

Evaluate AI model responses in Tamil and English for factual accuracy, instruction alignment, and helpfulness, providing ratings and justification to improve model training.

Annotated real and synthetic employee conversations for intents, topics, and sentiment, and performed quality control to ensure dataset accuracy for NLP models.

Compiled county-based real estate reports and sales data for California, delivering organized datasets and summaries for client use.

Tagged video features for machine learning algorithm development, ensuring accurate labels and temporal annotations to improve model performance.

Collected and analyzed e-commerce product data to support cataloging and quality improvement initiatives across product listings.

Created multilingual dialog datasets for NLP training, including annotations and metadata to support conversational AI development.

Extracted and categorized e-commerce product details from images to build labeled datasets for product recognition and recommendation systems.

Completed a Master of Science in Computer Science focusing on advanced computing topics and research-oriented coursework.

Completed a Bachelor of Science in Computer Science covering fundamental programming, algorithms, and systems concepts.