kshitij varshney profile photo 1 on Tal

kshitij varshney

Prompt Engineer & Evaluator @ Outlier | Prompt Engineering, LLM Evaluation

Greater Delhi Area

1+ year of work experience

Outlier
May 20262 mosPresent

Remote

India Managed Workforce (IMW) Program · Scale AI Design and evaluate agentic AI prompts for training large language models, ensuring multi-step tool use, skill discovery, and memory integration across complex synthetic environments. Build golden trajectories for autonomous AI agents by cross-referencing live database outputs with expected model behavior across health, finance, and personal management domains. Maintain quality standards across task submissions evaluated on correctness, completeness, efficiency, and naturality of AI model responses.
DataAnnotation
Oct 20259 mosPresent

Remote

• Annotated and evaluated datasets used for training and fine-tuning data for large language models (LLMs). • Performed quality checks on AI-generated outputs, including chatbot responses, reasoning accuracy, and content relevance. • Worked on diverse task types including data labeling, survey-based evaluation, and structured creative tasks following strict guidelines. • Applied domain knowledge to improve dataset accuracy and reduce model bias and hallucinations. • Collaborated asynchronously on multiple projects with evolving technical specifications. • Evaluated and compared multiple AI-generated outputs based on accuracy, safety, and usefulness.
Redrob
Sep 20252 mosOct 2025

Remote

Contributed to the development and optimization of AI datasets through precise and high-quality data annotation. Collaborated with a diverse team to ensure labeling consistency, accuracy, and efficiency in preparing training data for machine learning models. Worked closely with team leads to meet project goals, improve workflow processes, and maintain data integrity throughout the annotation lifecycle.
ULAVI TECHNOLOGIES PTE LTD
Jun 20243 mosAug 2024

Remote

I designed to significantly improve the trip-planning experience for users by integrating an AI-driven chat feature. The primary goal is to make the process of creating travel itineraries more personalized, interactive, and efficient. The AI will start by collecting detailed user preferences, including likes, dislikes, planned activities, and the number of people in the group. Based on this information, the AI will generate tailored itineraries that reflect the specific needs and interests of each user. Technologies Used: -Groq AI - used for generating smart, context-aware itineraries based on user inputs. -Swagger UI - used to create and document an API that connects the itinerary builder service with the existing chatbot. -Doubletick.io - the WhatsApp automation platform used to integrate the chatbot with users.

Education

amity University

Bachelor of Technology - B.Tech, computer science engineering

amity University

20214 yrs2025
Fairleigh Dickinson University

Bachelor of Technology - BTech, Computer Science engieering 3 continent program

Fairleigh Dickinson University

20242024

Join kshitij and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

kshitij varshney on Tal

Headline

Prompt Engineer & Evaluator @ Outlier | Prompt Engineering, LLM Evaluation

About

Currently contributing to Scale AI through prompt engineering and evaluation, focusing on designing agentic AI prompts and building golden trajectories for autonomous AI agents. Responsibilities include assessing large language model outputs for correctness, efficiency, and naturality, as well as ensuring robust memory integration and tool utilization in complex synthetic environments. Academic background in Computer Science Engineering spans institutions like Amity University, Fairleigh Dickinson University, and Birkbeck, University of London. Demonstrated interest in AI development and dataset annotation, with experience enhancing model accuracy and reducing bias. Motivated to explore roles in AI-powered product development and prompt engineering.

Location

Greater Delhi Area

Work experience

Prompt Engineer & Evaluator at Outlier, May 2026 - Present · 2 mos, Remote, India Managed Workforce (IMW) Program · Scale AI Design and evaluate agentic AI prompts for training large language models, ensuring multi-step tool use, skill discovery, and memory integration across complex synthetic environments. Build golden trajectories for autonomous AI agents by cross-referencing live database outputs with expected model behavior across health, finance, and personal management domains. Maintain quality standards across task submissions evaluated on correctness, completeness, efficiency, and naturality of AI model responses.

RLHF at DataAnnotation, Oct 2025 - Present · 9 mos, Remote, • Annotated and evaluated datasets used for training and fine-tuning data for large language models (LLMs). • Performed quality checks on AI-generated outputs, including chatbot responses, reasoning accuracy, and content relevance. • Worked on diverse task types including data labeling, survey-based evaluation, and structured creative tasks following strict guidelines. • Applied domain knowledge to improve dataset accuracy and reduce model bias and hallucinations. • Collaborated asynchronously on multiple projects with evolving technical specifications. • Evaluated and compared multiple AI-generated outputs based on accuracy, safety, and usefulness.

Data Annotator at Redrob, Sep 2025 - Oct 2025 · 2 mos, Remote, Contributed to the development and optimization of AI datasets through precise and high-quality data annotation. Collaborated with a diverse team to ensure labeling consistency, accuracy, and efficiency in preparing training data for machine learning models. Worked closely with team leads to meet project goals, improve workflow processes, and maintain data integrity throughout the annotation lifecycle.

Software Engineer at ULAVI TECHNOLOGIES PTE LTD, Jun 2024 - Aug 2024 · 3 mos, Remote, I designed to significantly improve the trip-planning experience for users by integrating an AI-driven chat feature. The primary goal is to make the process of creating travel itineraries more personalized, interactive, and efficient. The AI will start by collecting detailed user preferences, including likes, dislikes, planned activities, and the number of people in the group. Based on this information, the AI will generate tailored itineraries that reflect the specific needs and interests of each user. Technologies Used: -Groq AI - used for generating smart, context-aware itineraries based on user inputs. -Swagger UI - used to create and document an API that connects the itinerary builder service with the existing chatbot. -Doubletick.io - the WhatsApp automation platform used to integrate the chatbot with users.

Education

Bachelor of Technology - B.Tech, computer science engineering, amity University

Bachelor of Technology - BTech, Computer Science engieering 3 continent program, Fairleigh Dickinson University