Modassir Alam

Key Skills

Software

CVAT

Data Annotation Tech

Encord

iMerit

Labelbox

Scale AI

Telus

Top Subject Matter

No subject matter listed

Top Data Types

Text

Geospatial Tiled Imagery

Image

Video

3D Sensor

Top Task Types

Prompt + Response Writing (SFT)

Classification

Bounding Box

Polygon

Object Detection

Action Recognition

Tracking

Routing

Segmentation

Evaluation/Rating

Point/Key Point

Freelancer Overview

I am an AI/ML specialist with hands-on experience in large-scale data annotation, LLM training, and multimodal dataset creation for domains like computer vision, NLP, and autonomous driving. I have led annotation teams, designed QA workflows, and delivered high-quality labeled data using tools such as CVAT, Labelbox, and Label Studio. My expertise includes bounding box, polygon, and segmentation annotation, as well as prompt engineering and LLM output evaluation. I am skilled at optimizing annotation pipelines, refining taxonomies, and ensuring data consistency and accuracy through multi-layer QA and error analysis. My background in managing complex projects and developing production-grade datasets enables me to contribute to robust AI training pipelines and deliver reliable, scalable data solutions.

Labeling Experience

Data Annotation

Scale AIGeospatial Tiled ImageryEvaluation Rating

Evaluated and scored model outputs to improve conversational accuracy and response quality. • Conducted search evaluation including query–intent mapping and relevance scoring. • Annotated text data for intent classification, sentiment analysis, and short-form content. • Applied consistency checks to ensure high-quality, reliable training datasets.

Evaluated and scored model outputs to improve conversational accuracy and response quality. • Conducted search evaluation including query–intent mapping and relevance scoring. • Annotated text data for intent classification, sentiment analysis, and short-form content. • Applied consistency checks to ensure high-quality, reliable training datasets.

2024

OPS SME / AI/ML Associate - Data Annotation & AI Training

CVATImageSegmentation

Delivered high-quality image and video annotations for computer vision and multimodal training pipelines. Created and refined bounding box, polygon, segmentation, and keypoint annotations for autonomous vehicle datasets using CVAT, Labelbox, and Label Studio. Developed high-fidelity multi-turn conversational datasets and prompts for LLM alignment, significantly reducing hallucination rates. • Annotated vehicles, lanes, signals, pedestrians, and complex occlusions for AV projects. • Performed QA audits and taxonomy refinement for dataset scalability. • Managed multimodal content generation and prompt engineering. • Maintained 99%+ SLA adherence and large-scale team leadership.

Delivered high-quality image and video annotations for computer vision and multimodal training pipelines. Created and refined bounding box, polygon, segmentation, and keypoint annotations for autonomous vehicle datasets using CVAT, Labelbox, and Label Studio. Developed high-fidelity multi-turn conversational datasets and prompts for LLM alignment, significantly reducing hallucination rates. • Annotated vehicles, lanes, signals, pedestrians, and complex occlusions for AV projects. • Performed QA audits and taxonomy refinement for dataset scalability. • Managed multimodal content generation and prompt engineering. • Maintained 99%+ SLA adherence and large-scale team leadership.

2024

Multimodal Data Annotation

Scale AIImageBounding BoxPolygon

Delivered 10,000+ high-quality image and video annotations using multimodal pipelines.

Delivered 10,000+ high-quality image and video annotations using multimodal pipelines.

2024 - 2024

Autonomous Driving

Imerit3D SensorPoint Key Point

Annotated vehicles, lanes, traffic signals, and complex occlusions for production-grade AV datasets.

Annotated vehicles, lanes, traffic signals, and complex occlusions for production-grade AV datasets.

2024 - 2024

Action Refinement

EncordVideoObject DetectionAction Recognition

Annotating video datasets for autonomous action recognition using tools like Encord. It includes labeling human actions with bounding boxes, keypoints, and temporal annotations for training deep learning models.

Annotating video datasets for autonomous action recognition using tools like Encord. It includes labeling human actions with bounding boxes, keypoints, and temporal annotations for training deep learning models.

2024 - 2024

Show more

Education

Work History