Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Hi, I'm Vansh Goenka, an AI Engineer passionate about building intelligent solutions powered by large language models and computer vision. I enjoy developing efficient, reliable, and user-friendly AI applications and have experience working with Python, LangChain, and TensorFlow to create innovative AI agents and workflows. Currently, I focus on improving AI models through structured human feedback and data annotation, while also exploring projects that combine LLMs with real-time information retrieval. When I'm not coding, I contribute to the Google Developers Student Club and love connecting with others who share my enthusiasm for AI.…Hi, I'm Vansh Goenka, an AI Engineer passionate about building intelligent solutions powered by large language models and computer vision. I enjoy developing efficient, reliable, and user-friendly AI applications and have experience working with Python, LangChain, and TensorFlow to create innovative AI agents and workflows. Currently, I focus on improving AI models through structured human feedback and data annotation, while also exploring projects that combine LLMs with real-time information retrieval. When I'm not coding, I contribute to the Google Developers Student Club and love connecting with others who share my enthusiasm for AI.

Vansh Goenka

Programmer, AI Engineer, AI Developer, +5





Hi, I'm Vansh Goenka, an AI Engineer passionate about building intelligent solutions powered by large language models and computer vision. I enjoy developing efficient, reliable, and user-friendly AI applications and have experience working with Python, LangChain, and TensorFlow to create innovative AI agents and workflows. Currently, I focus on improving AI models through structured human feedback and data annotation, while also exploring projects that combine LLMs with real-time information retrieval. When I'm not coding, I contribute to the Google Developers Student Club and love connecting with others who share my enthusiasm for AI.…Hi, I'm Vansh Goenka, an AI Engineer passionate about building intelligent solutions powered by large language models and computer vision. I enjoy developing efficient, reliable, and user-friendly AI applications and have experience working with Python, LangChain, and TensorFlow to create innovative AI agents and workflows. Currently, I focus on improving AI models through structured human feedback and data annotation, while also exploring projects that combine LLMs with real-time information retrieval. When I'm not coding, I contribute to the Google Developers Student Club and love connecting with others who share my enthusiasm for AI.

Available to hire

Hi, I’m Vansh Goenka, an AI Engineer passionate about building intelligent solutions powered by large language models and computer vision. I enjoy developing efficient, reliable, and user-friendly AI applications and have experience working with Python, LangChain, and TensorFlow to create innovative AI agents and workflows.

Currently, I focus on improving AI models through structured human feedback and data annotation, while also exploring projects that combine LLMs with real-time information retrieval. When I’m not coding, I contribute to the Google Developers Student Club and love connecting with others who share my enthusiasm for AI.

Skills

Experience Level

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Language

English

Fluent

Work Experience

Software Engineer for Training AI Data (RLHF) at G2i

March 1, 2025 - Present

Improved Large Language Models (LLM) coding performance by testing outputs, identifying issues, and aligning responses with prompt intent. Designed and iteratively refined prompts for LLMs, contributing to performance tuning, instruction-following, behavior and inference efficiency using Python. Edited LLM-generated code responses to enhance clarity, correctness, and alignment with prompt intent.

AI Intern at CodeSpaze

December 31, 2024 - July 22, 2025

Implemented advanced computer vision techniques through CNN architectures trained on 2D image data for efficient object detection in varied lighting and occlusion. Enhanced model accuracy from 84% to almost 90% using ensemble techniques like simple averaging and majority voting. Developed a Streamlit web application for real-time image uploads and live classification results.

Software Engineer for Training AI Data (RLHF) at G2i

March 1, 2025 - Present

Improve Large Language Models (LLM) coding performance by testing outputs, identifying issues, and aligning responses with prompt intent. Designed and iteratively refined prompts for LLMs, contributing to performance tuning, instruction-following, behavior, and inference efficiency using Python. Edited LLM-generated code responses to enhance clarity, correctness, and alignment with prompt intent.

AI Intern at CodeSpaze

December 31, 2024 - July 22, 2025

Implemented advanced computer vision techniques through CNN architectures trained on 2D image data, for efficient object detection in varied lighting and occlusion. Enhanced model accuracy from 84% to almost 90% using ensemble techniques like simple averaging and majority voting. Developed a Streamlit web application for real-time image uploads and live classification results.

Software Engineer for Training AI Data (RLHF) at G2i

March 1, 2025 - Present

Analyzed hundreds of LLM responses per project, scoring them against key criteria such as factual accuracy, prompt adherence, and clarity to generate high-quality training data. Authored corrected, high-fidelity responses for underperforming outputs to directly improve model performance via the reinforcement learning loop. Contributed to a 12% increase in LLM output accuracy and safety by providing structured human feedback and detailed data annotation.

AI Intern at CodeSpaze

December 31, 2024 - July 22, 2025

Implemented advanced computer vision techniques through CNN architectures trained on 2D image data, for efficient image processing in varied lighting and occlusion. Boosted model accuracy from 84% to 90% by applying ensemble strategies like simple averaging and majority voting. Deployed classification models via a Streamlit web application, enabling real-time image uploads and live inference results.