Smitkumar Patel

Key Skills

Software

OpenCV AI Kit (OAK)

Label Studio

Roboflow

Top Subject Matter

Traffic management

vehicle detection

computer vision

Top Data Types

Video

Image

Computer Code Programming

Top Task Types

Object Detection

Bounding Box

Question Answering

Text Generation

Text Summarization

Computer Programming/Coding

Freelancer Overview

I have hands-on experience working with AI systems through multiple real-world projects involving data processing, annotation, and model integration. I have built end-to-end AI applications such as a Video Question Answering System, where I handled audio transcription using Whisper, extracted and processed video frames using OpenCV, and structured the data for retrieval using FAISS-based vector databases. I also developed a PDF Question Answering system using LangChain and embeddings, which required cleaning, chunking, and organizing textual data—similar to data labeling workflows used in training AI models. In addition, I have practical exposure to computer vision tasks like vehicle detection using YOLO, where I worked with image/video data and bounding box-based understanding. My strong foundation in Python, data preprocessing (NumPy, Pandas), and AI pipelines allows me to understand how high-quality annotated data improves model performance. I am detail-oriented, capable of handling large datasets, and focused on accuracy and consistency—key qualities required in AI training and data annotation roles.

Labeling Experience

Machine Learning Intern: Traffic Management Data Labeling

Opencv AI Kit OakImageObject Detection

I developed a real-time traffic management system focused on vehicle detection, data labeling, and analytics using Python, OpenCV, and YOLOv8. My primary responsibility included building an end-to-end pipeline for processing traffic video data, where I automated object detection and generated structured labeled datasets for multi-lane vehicle analysis. I worked extensively with image/video annotation workflows, leveraging YOLOv8 for bounding box-based object detection and validation of labeled outputs. I ensured high-quality annotation by optimizing detection accuracy and handling edge cases such as occlusions, varying lighting conditions, and dense traffic scenarios. • Utilized YOLOv8 for automated object detection and annotation of vehicles in video datasets • Processed and structured labeled data with timestamps for analytics and model training • Designed SQL-based storage for efficient management of annotated vehicle data • Improved labeling accuracy through validation and preprocessing techniques (>90% accuracy achieved) • Applied computer vision techniques for multi-lane vehicle counting and traffic pattern analysis

I developed a real-time traffic management system focused on vehicle detection, data labeling, and analytics using Python, OpenCV, and YOLOv8. My primary responsibility included building an end-to-end pipeline for processing traffic video data, where I automated object detection and generated structured labeled datasets for multi-lane vehicle analysis. I worked extensively with image/video annotation workflows, leveraging YOLOv8 for bounding box-based object detection and validation of labeled outputs. I ensured high-quality annotation by optimizing detection accuracy and handling edge cases such as occlusions, varying lighting conditions, and dense traffic scenarios. • Utilized YOLOv8 for automated object detection and annotation of vehicles in video datasets • Processed and structured labeled data with timestamps for analytics and model training • Designed SQL-based storage for efficient management of annotated vehicle data • Improved labeling accuracy through validation and preprocessing techniques (>90% accuracy achieved) • Applied computer vision techniques for multi-lane vehicle counting and traffic pattern analysis

2025 - 2025

Education

Work History