Prafful P.

Key Skills

Software

No software listed

Top Subject Matter

Tractor valuation and vehicle assessments

Multilingual product recommendation and search systems

Telecom infrastructure digitization and survey automation

Top Data Types

Image

Text

Document

Top Task Types

Object Detection

Prompt + Response Writing (SFT)

Bounding Box

Entity (NER) Classification

Segmentation

Freelancer Overview

Data Labeling and AI Training for Tractor Valuation System. Brings 7+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include YOLO, Internal, and Proprietary Tooling. Education includes Bachelor of Technology, Delhi Technological University (2019) and Senior Secondary Education, Kendriya Vidyalaya AFS Chakeri Kanpur (2015). AI-training focus includes data types such as Image, Text, and Document and labeling workflows including Object Detection, Prompt + Response Writing (SFT), and Bounding Box.

Labeling Experience

LLM Prompt/Response Labeling for Product Assistant

TextPrompt + Response Writing (SFT)

Architected a multilingual product search assistant utilizing LLM and RAG for labeled prompt/response data. Enabled support for both text and audio input modalities, expanding accessibility for product retrieval. Labeled datasets were generated for product search queries and managed for multiple Indian languages. • Designed dual-prompt guardrail system to enhance response quality. • Annotated multilingual queries to train LLM models. • Integrated and semantically labeled data for RAG pipelines. • Produced structured outputs for system evaluation and training.

2025 - 2025

Data Labeling and AI Training for Tractor Valuation System

ImageObject Detection

Built a multimodal AI pipeline combining YOLO, GPT-4 Vision, and Gemini 2.5 Vision to extract structured insights from tractor and RC images. Designed workflows for metadata validation, depreciation scoring, and automated valuation summary generation using labeled visual data. Developed and deployed image annotation models for real-world tractor valuation scenarios. • Combined computer vision with LLM-based reasoning for data extraction. • Labeled images to identify tractor features such as tyre wear, rust, and attachments. • Implemented models to automate object detection and classification. • Utilized YOLO and LLM tools for structured insight extraction.

2025 - 2025

Document Annotation and Object Detection for Form Digitalization

DocumentObject Detection

Developed digitalization workflows for converting PDF forms to enhanced document images and labeling handwritten text for extraction. Labeled checkboxes and mapped their positions using edge detection for structured form completion. Produced high-accuracy labeled datasets for model development and evaluation. • Labeled forms for field/checkbox detection in random layouts. • Annotated handwritten text for entity extraction models. • Created ground truth data for classification and extraction. • Supported automation of digital form completion with labeled images.

2019 - 2025

Satellite Image Segmentation and Feature Labeling

ImageSegmentation

Trained YOLO and UNet models to segment and extract features from satellite images, focusing on transport-related attributes. Labeled road features such as bus lanes, cycle lanes, and road markings for geospatial AI processing. Segmented large images into tiled datasets to streamline the annotation workflow. • Annotated satellite imagery for fine-grained land use features. • Used segmentation models to identify multiple classes in geospatial data. • Enhanced automation pipelines with labeled high-resolution tiles. • Applied feature extraction methods to improve processing speed.

2019 - 2025

Document Labeling for Easement Polygon Creation

DocumentEntity (NER) Classification

Performed entity extraction and classification from Ordinance Survey PDF documents as part of easement polygon automation. Processed and labeled document text and spatial data after OCR transformation. Labeled features enabled automated geometric drawing in AutoCAD systems. • Converted survey PDFs to structured images for OCR processing. • Labeled extracted text entities linked to geospatial features. • Automated mapping of legal property boundaries. • Supported pipeline integration with AutoCAD for feature rendering.

2019 - 2025

Show more

Education

Work History