Remote Active-Learning ML Engineer for Computer Vision

Blue Signal Search; United States; On-site; USD 120,000 - 190,000; Full time

Benefits offered by this job

Equity participation

Remote work flexibility

Ownership of data platform

Job summary

Blue Signal Search is seeking an Active Learning ML Engineer to build scalable pipelines that convert large volumes of raw video into dependable training data for computer vision models. This role offers remote work nationwide with EST/CST hours, and significant ownership over a core ML data platform.

You will work across software, data, and ML, applying state-of-the-art foundation models and modern cloud tools to accelerate labeling, validation, and model improvement while maintaining high

Qualifications

• Production ML infrastructure experience
• Strong Python and software engineering discipline
• Experience with data pipelines and ML training data pipelines

Responsibilities

• Build scalable pipelines transforming raw video into organized training datasets
• Create active learning and weak supervision workflows for labeling and review
• Apply foundation models (e.g., Grounding DINO, SAM 2) to accelerate labeling
• Develop internal apps for annotator validation and workflow automation
• Own data quality across cloud ingestion, cleaning, organization and delivery of training assets
• Establish evaluation datasets and quality controls for ML performance

Skills

Python

FFmpeg

SQL

Computer vision

Active learning

Foundation models

AWS/GCP

PyTorch/TensorFlow

Web tooling

Tools

React

Next.js

FastAPI

Flask

SageMaker

Job description

Blue Signal Search is seeking an Active Learning ML Engineer to build scalable pipelines that convert large volumes of raw video into dependable training data for computer vision models. This role offers remote work nationwide with EST/CST hours, and significant ownership over a core ML data platform.

You will work across software, data, and ML, applying state-of-the-art foundation models and modern cloud tools to accelerate labeling, validation, and model improvement while maintaining high