For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
E
Enoch

Enoch

Agency
India flagHyderabad, India
www.dexset.ai

Key Skills

Software

No software listed

Top Subject Matter

No subject matter listed

Top Data Types

VideoVideo
ImageImage
3D Sensor3D Sensor

Top Task Types

Bounding BoxBounding Box
SegmentationSegmentation
Point/Key PointPoint/Key Point
CuboidCuboid
Object DetectionObject Detection

Company Overview

Robots don't learn from clean theory; they learn from messy reality. Dexset helps robotics teams train physical AI with data from real environments. We collect, annotate, structure, and deliver multimodal datasets for manipulation, navigation, human-object interaction, and teleoperation. Our workflow captures synchronized egocentric (first-person) and exocentric (third-person) video, plus the unpredictable edge cases simulation misses—lighting changes, occlusions, novel objects, and failure cases. Expert annotation teams label interactions, grasp points, keypoints, 3D bounding boxes, and temporal sequences with rigorous human-in-the-loop quality control. We deliver task-level annotated, ML-ready data in standard formats (COCO, YOLO, JSON) directly into your pipeline via API, S3, or cloud storage. Fully compliant with GDPR and CCPA. Based in Hyderabad, India with a 50+ person workforce, Dexset partners with AI labs and robotics teams to improve grasp success rates, cut triage time, and accelerate autonomous task learning.

Security

Security Overview

Dexset operates with strict data security and privacy controls suited to robotics and physical-AI datasets. All projects are governed by signed NDAs and data processing agreements, with role-based access controls limiting dataset access to assigned, vetted annotators. Data is stored and transferred over encrypted channels (TLS in transit; encryption at rest) and delivered securely via API, S3, or client cloud storage. Personally identifiable information is minimized and, where required, faces/license plates and other sensitive elements are blurred or redacted. Our workflows are fully compliant with GDPR and CCPA, and we support client-specific data handling, retention, and deletion requirements. Quality and integrity are maintained through human-in-the-loop QC with multi-stage review. Security training and confidentiality obligations apply to our full 50+ person workforce based in Hyderabad, India.

Security Credentials

GDPR

Labeling Experience

Robotic Manipulation Video Annotation for Imitation Learning

Internal/Proprietary ToolingVideoVideoCuboidCuboidSegmentationSegmentation

Ongoing video-annotation program for robotic manipulation and imitation-learning datasets. Work includes frame-accurate object and gripper tracking, 3D cuboid annotation of manipulated objects, semantic segmentation of scenes and target objects, and key-point/pose labeling of robot end-effectors and human-demonstration hands across multi-view video. Built for training and evaluating manipulation and grasping policies. Quality is maintained through multi-pass human-in-the-loop QC, gold-standard benchmark tasks, consensus review, and per-batch acceptance thresholds.

2025 - 2026