Akshat Jain

Key Skills

Software

Remotasks

Scale AI

Other

Top Subject Matter

Computer Programming/Coding

Databases and SQL

Evaluation

Top Data Types

Computer Code Programming

Document

Image

Top Task Types

Computer Programming/Coding

Evaluation/Rating

Function Calling

Prompt + Response Writing (SFT)

RLHF

Freelancer Overview

In data labeling and AI training data, my expertise is anchored in crafting intricate prompts that challenge AI models, ensuring their resilience and accuracy across diverse domains. I specialize in categorizing and annotating complex data science tasks like data visualization, manipulation, cleaning, feature engineering, statistical analysis, and SQL query optimization. My approach involves adversarial prompt generation to systematically expose model vulnerabilities, refining prompts iteratively until they pinpoint failure points. Additionally, I have led SQL projects that integrate advanced query design and database optimization techniques, contributing to comprehensive data handling and analysis solutions. Through multi-model evaluations, I assess and enhance model performance, leveraging dynamic prompt generation algorithms to adapt strategies based on real-time feedback. These experiences underscore my ability to deliver high-quality AI training data and drive continuous improvement in model training and deployment strategies.

Labeling Experience

Challenging AI Models through Complex Data Science Prompts: Adversarial Generation and Multi-Model Evaluation

Scale AIComputer Code ProgrammingPrompt Response Writing SFT

Scope: The project involves generating complex data science prompts to challenge AI models until they fail. Categories include data visualization, manipulation, cleaning, feature engineering, matrix operations, statistical analysis, and model fine-tuning. Prompts are refined until they are specific and difficult enough to cause model errors. Data Labelling: Annotations cover data visualization, manipulation, cleaning, feature engineering, matrix operations, statistical analysis, and model fine-tuning, ensuring precise and accurate prompts. Size and Complexity: Large-scale project with numerous complex prompts across various data science categories, requiring advanced understanding and detailed annotations. Quality Measures: Ensures specificity and relevance of prompts, validates their complexity, continuous refinement based on model performance, peer reviews, automated checks, and regular feedback loops to maintain high standards.

Scope: The project involves generating complex data science prompts to challenge AI models until they fail. Categories include data visualization, manipulation, cleaning, feature engineering, matrix operations, statistical analysis, and model fine-tuning. Prompts are refined until they are specific and difficult enough to cause model errors. Data Labelling: Annotations cover data visualization, manipulation, cleaning, feature engineering, matrix operations, statistical analysis, and model fine-tuning, ensuring precise and accurate prompts. Size and Complexity: Large-scale project with numerous complex prompts across various data science categories, requiring advanced understanding and detailed annotations. Quality Measures: Ensures specificity and relevance of prompts, validates their complexity, continuous refinement based on model performance, peer reviews, automated checks, and regular feedback loops to maintain high standards.

2024 - 2024

Multi-Language Query Generation

Scale AIComputer Code ProgrammingPrompt Response Writing SFT

Scope: The project involves generating prompts to fetch data from Google Cloud Platform based on given database details, query types, and sub-query types. Responses are written in BigQuery, SQL, or PostgreSQL, followed by converting the response to another language from the set. Data Labelling: Annotated data includes a variety of query and sub-query types for Google Cloud Platform databases, with detailed labelling for implementations in BigQuery, SQL, and PostgreSQL. Size and Complexity: Extensive project involving large-scale data and complex query types, with comprehensive annotation and multi-language code conversion. Quality Measures: Ensures prompt accuracy, query consistency, response correctness, and efficient code conversion. Quality is maintained through peer reviews, automated validations, and continuous feedback.

Scope: The project involves generating prompts to fetch data from Google Cloud Platform based on given database details, query types, and sub-query types. Responses are written in BigQuery, SQL, or PostgreSQL, followed by converting the response to another language from the set. Data Labelling: Annotated data includes a variety of query and sub-query types for Google Cloud Platform databases, with detailed labelling for implementations in BigQuery, SQL, and PostgreSQL. Size and Complexity: Extensive project involving large-scale data and complex query types, with comprehensive annotation and multi-language code conversion. Quality Measures: Ensures prompt accuracy, query consistency, response correctness, and efficient code conversion. Quality is maintained through peer reviews, automated validations, and continuous feedback.

2024 - 2024

Iterative Prompt Generation and Evaluation

Scale AIComputer Code ProgrammingEvaluation Rating

Scope: The project entails an advanced system for generating prompts and conducting comparative evaluations of two responses in real-time. Following the selection of the optimal response, the system iteratively refines the prompt to elicit deeper and more nuanced responses Data Labelling: Annotated data includes Python libraries crucial to data science (pandas, numpy, scikit-learn, matplotlib, seaborn) for precise response generation. Size and Complexity: Moderate scale with emphasis on Python libraries, demanding advanced algorithms and model architectures. Quality Measures: Ensures prompt clarity, response relevance, adherence to criteria, depth of information, and contextual appropriateness for high-quality outputs.

Scope: The project entails an advanced system for generating prompts and conducting comparative evaluations of two responses in real-time. Following the selection of the optimal response, the system iteratively refines the prompt to elicit deeper and more nuanced responses Data Labelling: Annotated data includes Python libraries crucial to data science (pandas, numpy, scikit-learn, matplotlib, seaborn) for precise response generation. Size and Complexity: Moderate scale with emphasis on Python libraries, demanding advanced algorithms and model architectures. Quality Measures: Ensures prompt clarity, response relevance, adherence to criteria, depth of information, and contextual appropriateness for high-quality outputs.

2024 - 2024

Education

Work History