Vishnu Iyengar

Key Skills

Software

Scale AI

CVAT

Appen

Top Subject Matter

No subject matter listed

Top Data Types

Text

Image

Audio

Top Label Types

RLHF

Evaluation Rating

Prompt Response Writing SFT

Polygon

Segmentation

Classification

Object Detection

Emotion Recognition

Audio Recording

Transcription

Freelancer Overview

I am an AI Training Specialist and Data Scientist with over 1-2 years of experience in high-fidelity data curation and model alignment. My expertise lies in bridging the gap between raw datasets and production-ready AI through Reinforcement Learning from Human Feedback (RLHF), Supervised Fine-Tuning (SFT), and high-precision Computer Vision annotation. I have a proven track record of developing "Gold Standard" datasets for frontier LLMs and RAG-based systems, specifically focusing on hallucination mitigation, multi-step logical reasoning, and complex technical evaluation across healthcare, real estate, and e-commerce domains. Beyond standard labeling, I specialize in QA/QC architectural design and the resolution of complex edge cases in multi-modal data. Whether performing pixel semantic segmentation for autonomous systems or phonetic labeling for ASR models, I prioritize data integrity as the primary lever for model performance. I am proficient in leveraging tools like Scale AI, CVAT, and Labelbox alongside technical frameworks like PyTorch, Hugging Face, and LangChain to automate preprocessing and evaluate model output against rigorous human-centric benchmarks.

Labeling Experience

RLHF Advanced Technical Reasoning

Scale AITextRLHFEvaluation Rating

Contributed to alignment and fine-tuning of a frontier Large Language Model. Focused on Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF) to improve model performance in complex, multi-step reasoning tasks. Performed ranking and preference labeling on model outputs to minimize hallucinations and ensure technical factualness.

Contributed to alignment and fine-tuning of a frontier Large Language Model. Focused on Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF) to improve model performance in complex, multi-step reasoning tasks. Performed ranking and preference labeling on model outputs to minimize hallucinations and ensure technical factualness.

2024 - 2025

Multi-dialect Speech Transcription

AppenAudioEmotion RecognitionEvaluation Rating

Annotated and transcribed high-fidelity audio datasets to train Automatic Speech Recognition (ASR) models. Focused on labeling, identifying nuances in tone, pitch, and cadence to improve the naturalness of AI-generated speech. Evaluated model outputs for accuracy and corrected errors in accent-heavy or noisy environments.

Annotated and transcribed high-fidelity audio datasets to train Automatic Speech Recognition (ASR) models. Focused on labeling, identifying nuances in tone, pitch, and cadence to improve the naturalness of AI-generated speech. Evaluated model outputs for accuracy and corrected errors in accent-heavy or noisy environments.

2023 - 2025

Instance Segmentation

CVATImagePolygonSegmentation

Managed high-precision image annotation for large-scale computer vision dataset. Specialized in pixel-level segmentation and polygon-based instance annotation for complex urban environments. My role required identifying and labeling overlapping objects and defining precise boundaries in low-visibility imagery

Managed high-precision image annotation for large-scale computer vision dataset. Specialized in pixel-level segmentation and polygon-based instance annotation for complex urban environments. My role required identifying and labeling overlapping objects and defining precise boundaries in low-visibility imagery

2023 - 2025

Education

Work History