Somil Verma

Key Skills

Software

CrowdSource

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Top Task Types

Entity Ner Classification

Question Answering

Text Generation

Text Summarization

Freelancer Overview

I have hands-on experience in training and evaluating machine learning models, with a strong focus on data quality and performance analysis. At Outlier AI, I worked as a Model Trainer where I analyzed large datasets using Python and SQL, identified gaps in model outputs, and refined prompts to improve overall response quality by 20%. My background in backend and full-stack development, along with projects in domains like medical appointment systems and e-commerce platforms, has given me a solid understanding of data structures, database design, and the importance of accurate data annotation. I am skilled in tools such as Python, SQL, MongoDB, and Git, and I am committed to ensuring high-quality, well-labeled data for AI training workflows.

Labeling Experience

Prompt Engineering and Evaluation – Outlier AI

CrowdsourceComputer Code ProgrammingEntity Ner ClassificationQuestion Answering

Worked as a Prompt Engineering Intern on a large-scale LLM training and evaluation project. Contributed to various Natural Language Processing (NLP) tasks such as prompt creation, revision, and completion generation for reasoning, coding, and conversational objectives. Responsible for designing high-quality inputs for instruction-following models, reviewing AI outputs, and generating unit test cases for coding problems. Tasks included evaluating completions, fine-tuning prompts, and ensuring alignment with annotation guidelines. Maintained high accuracy and consistency across thousands of examples. The project adhered to strict quality assurance protocols and targeted SOTA model performance improvement.

Worked as a Prompt Engineering Intern on a large-scale LLM training and evaluation project. Contributed to various Natural Language Processing (NLP) tasks such as prompt creation, revision, and completion generation for reasoning, coding, and conversational objectives. Responsible for designing high-quality inputs for instruction-following models, reviewing AI outputs, and generating unit test cases for coding problems. Tasks included evaluating completions, fine-tuning prompts, and ensuring alignment with annotation guidelines. Maintained high accuracy and consistency across thousands of examples. The project adhered to strict quality assurance protocols and targeted SOTA model performance improvement.

2024 - 2024

Education

Work History