Pranav R.

Key Skills

Software

Scale AI

Other

Don't disclose

Top Subject Matter

Large language model (LLM) training and prompting for request parsing and response generation

Computer vision model training for freight image quality and defect detection

Compliance operations with automated data validation and quality evaluation

Top Data Types

Image

Document

Text

Top Task Types

Prompt + Response Writing (SFT)

Object Detection

Freelancer Overview

Freelance Data Scientist – AI Training (Scale AI / Outlier). Brings 7+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Scale AI, Other, and Don't disclose. Education includes Master of Science, University of Chicago (2023) and Bachelor of Technology, CMR University (2020). AI-training focus includes data types such as Computer Code, Programming, and Image and labeling workflows including Prompt + Response Writing (SFT), Object Detection, and Evaluation.

Labeling Experience

Freelance Data Scientist – AI Training (Scale AI / Outlier)

Scale AIPrompt + Response Writing (SFT)

Performed AI training and evaluation work by debugging LLM behavior and improving parsing performance. Used Chain-of-Thought prompting to train Gemini with multi-step reasoning, optimizing response quality and efficiency. Measured impact through reductions in response error rate and redundant API calls while improving parsing accuracy and overall response efficiency. • Debugging framework for LLM inefficiencies and logic errors • Chain-of-Thought prompting for Gemini multi-step reasoning • Parsing of user requests to improve downstream accuracy • Evaluation metrics: error-rate reduction, redundancy reduction, efficiency and accuracy gains

2024 - Present

Data Scientist (Capstone Partner) – vision model training for defect identification

OtherImageObject Detection

Conducted AI training/data labeling-oriented model development using computer vision to identify poor-quality freight images. Analyzed a large image set with transfer learning and YOLO to detect issues such as poorly loaded freight and odd freight types. Built and trained multiple convolutional neural networks to predict image quality and freight-related defects, achieving strong predictive accuracy. • Image analysis on 6,500 freight images using YOLO + transfer learning • Detection of poor-quality images, poorly loaded freight, and odd freight types • ETL pipeline development to reduce preparation time for model training • CNN model development for predicting dark, blurry, and odd-shaped freight

2023 - 2023

Senior Associate, PICCCE, Compliance Operations (Amazon Development Centre)

Don't discloseDocument

Implemented automated compliance checks by validating data against region-specific standards on Amazon’s marketplace. Improved detection quality for faulty products through analysis and operational study, increasing true positives and reducing customer complaints. Built automated reporting to analyze non-compliant selling partners and support compliance operations. • Data validation scripts for automated compliance checks • Time-and-motion study to improve identification of faulty products • Automated compliance reporting with Power BI/DAX • SQL development to extract compliance-related data and support investigations

2020 - 2022

Education

Work History