Arjun Vettickal Anilkumar

Key Skills

Software

Label Studio

Roboflow

Top Subject Matter

No subject matter listed

Top Data Types

Image

Top Label Types

Bounding Box

Object Detection

Polygon

Segmentation

Freelancer Overview

I specialize in synthetic data generation and annotation for AI and robotics applications, with hands-on experience developing pipelines for object detection and benchmarking synthetic data workflows using tools like NVIDIA Omniverse Replicator and Blender. My background includes automating robotic simulation workflows for industrial metaverse platforms. I have strong programming skills in Python, C++, and C#, and have led projects involving generative AI, CAD modeling, and VR-based training environments. My technical expertise help me to create and optimise high-quality annotated test sets for bin-picking tasks, ensuring robust AI model performance in real-world scenarios.

Labeling Experience

Technical Student Assistant — Data Annotation/QA (ARENA2036 e.V)

RoboflowImageBounding Box

Performed 2D annotation and quality assurance for computer vision datasets using established guidelines. Managed consistent class mapping, occlusion, and truncation rules to maintain data quality throughout annotations. Exported labels in YOLO and COCO formats for use in machine learning workflows. • Conducted bounding box labeling for datasets featuring industrial parts and CAD assets. • Utilized Roboflow for annotation, export, and version control with Python scripts for validation and automation. • Conducted second-pass reviews and spot checks to ensure accurate annotations were delivered. • Ensured train/val splits, stable class IDs, and strict naming conventions.

Performed 2D annotation and quality assurance for computer vision datasets using established guidelines. Managed consistent class mapping, occlusion, and truncation rules to maintain data quality throughout annotations. Exported labels in YOLO and COCO formats for use in machine learning workflows. • Conducted bounding box labeling for datasets featuring industrial parts and CAD assets. • Utilized Roboflow for annotation, export, and version control with Python scripts for validation and automation. • Conducted second-pass reviews and spot checks to ensure accurate annotations were delivered. • Ensured train/val splits, stable class IDs, and strict naming conventions.

2025

Master's Thesis — Benchmarking Synthetic Data Generation (Annotation/QA)

RoboflowImageBounding Box

Developed and benchmarked synthetic data generation pipelines for object detection. Standardized and unified annotations across real and synthetic datasets using clear rules and processes. Ensured data quality with second-pass reviews, range checks, and train/val split management. • Utilized Roboflow to annotate and QA real images for consistency with synthetic data benchmarks. • Converted COCO to YOLO formats while enforcing stable class mapping across all experiments. • Implemented deterministic label exports and overlays for precise validation. • Focused on minimizing label drift and maximizing train/val stability.

Developed and benchmarked synthetic data generation pipelines for object detection. Standardized and unified annotations across real and synthetic datasets using clear rules and processes. Ensured data quality with second-pass reviews, range checks, and train/val split management. • Utilized Roboflow to annotate and QA real images for consistency with synthetic data benchmarks. • Converted COCO to YOLO formats while enforcing stable class mapping across all experiments. • Implemented deterministic label exports and overlays for precise validation. • Focused on minimizing label drift and maximizing train/val stability.

2022

Education

Work History