Cliff Masira

Key Skills

Software

Label Studio

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Top Label Types

Geocoding

Mapping

Computer Programming Coding

Data Collection

Prompt Response Writing SFT

Freelancer Overview

I am an experienced AI Trainer and Machine Learning Specialist with a strong background in data labeling, annotation strategy, and quality assurance for AI and machine learning projects. My expertise spans training and fine-tuning large language models (LLMs), implementing reinforcement learning with human feedback (RLHF), and developing structured annotation guidelines for teams of 50+ annotators. I have led projects in NLP, computer vision, and generative AI, building scalable data pipelines and automated quality control systems that consistently achieve over 98% labeling accuracy. My technical toolkit includes Python, PyTorch, TensorFlow, Hugging Face, SQL, and cloud platforms like AWS and GCP. I am passionate about bridging technical development with business needs, ensuring data quality, model fairness, and high-impact AI deployment across domains such as chatbots, content moderation, and defect detection.

Labeling Experience

Image Labeling/Text Labeling

Label StudioComputer Code ProgrammingGeocodingMapping

1️⃣ Scope of the Project The project focused on preparing high-quality training data for perception models used in autonomous driving systems. The objective was to improve real-time object detection, lane understanding, depth estimation, and scene interpretation under diverse environmental conditions. Core goals: Enhance object detection accuracy (vehicles, pedestrians, cyclists, traffic signs) Improve lane boundary and drivable area segmentation Strengthen 3D object tracking using LiDAR data Reduce false positives in complex urban scenarios Support model generalization across weather, lighting, and geographic variations The dataset supported deep learning models based on CNNs and transformer-based perception architectures deployed in advanced driver-assistance systems (ADAS) and fully autonomous stacks. 2️⃣ Specific Data Labeling Tasks Performed 🔹 Image Annotation 2D bounding boxes for vehicles, pedestrians, cyclists, traffic lights Semantic segmentation (road, sidewalk,

1️⃣ Scope of the Project The project focused on preparing high-quality training data for perception models used in autonomous driving systems. The objective was to improve real-time object detection, lane understanding, depth estimation, and scene interpretation under diverse environmental conditions. Core goals: Enhance object detection accuracy (vehicles, pedestrians, cyclists, traffic signs) Improve lane boundary and drivable area segmentation Strengthen 3D object tracking using LiDAR data Reduce false positives in complex urban scenarios Support model generalization across weather, lighting, and geographic variations The dataset supported deep learning models based on CNNs and transformer-based perception architectures deployed in advanced driver-assistance systems (ADAS) and fully autonomous stacks. 2️⃣ Specific Data Labeling Tasks Performed 🔹 Image Annotation 2D bounding boxes for vehicles, pedestrians, cyclists, traffic lights Semantic segmentation (road, sidewalk,

2023

AI annotator

LabelboxVideoBounding BoxRelationship

Labeled 80K+ video clips for action recognition and event detection using custom ML-assisted workflows. Configured labeling interfaces with dynamic taxonomies and conditional logic. Increased model training accuracy by 15% through high-consistency temporal tagging and structured QA audits.

Labeled 80K+ video clips for action recognition and event detection using custom ML-assisted workflows. Configured labeling interfaces with dynamic taxonomies and conditional logic. Increased model training accuracy by 15% through high-consistency temporal tagging and structured QA audits.

2022 - 2022

Education

Work History