AI/ML Computer Vision Engineer — YOLO Specialist (Hybrid VA)

Socket.dev; Ashburn (VA); On-site; USD 87,000 - 182,000; Full time

Job summary

CAC I is seeking a Mid-Level AI/ML Computer Vision Engineer in Northern Virginia to join the BEAGLE ASF team. You will train YOLO models, curate datasets, and build end-to-end inference pipelines for live FMV analysis.

You’ll work with a small pod and senior staff in an Agile setting to deliver production-ready code. The role requires 2+ years of hands-on CV model experience, strong Python and ML library skills, OpenCV/FFmpeg, and a focus on model performance metrics.

Qualifications

• U.S. Citizen with ability to pass CBP background check and related screenings.
• Bachelor’s degree in CS, DS, SE, or related discipline required.
• 3–5 years in software engineering or data science with 2+ years building CV models.
• Hybrid work availability in Ashburn, VA.

Responsibilities

• Train, fine-tune and evaluate YOLO object-detection models on proprietary data.
• Process live/ recorded video to build high-quality training datasets.
• Annotate and curate datasets, generate synthetic data when needed.
• Develop inference pipelines to export structured predictions (JSON/Protobuf).
• Produce reports with metrics (mAP, FPS, latency) to improve models.
• Collaborate in an Agile/Scrum team to deliver production-ready code.

Skills

Python

YOLO development

PyTorch/TensorFlow

Model evaluation

Benchmarking

Data preprocessing

Git

Education

B.S. in Computer Science / Data Science / Software Engineering

Tools

OpenCV

FFmpeg

CVAT/LabelImg

Git

Job description

CAC I is seeking a Mid-Level AI/ML Computer Vision Engineer in Northern Virginia to join the BEAGLE ASF team. You will train YOLO models, curate datasets, and build end-to-end inference pipelines for live FMV analysis.

You’ll work with a small pod and senior staff in an Agile setting to deliver production-ready code. The role requires 2+ years of hands-on CV model experience, strong Python and ML library skills, OpenCV/FFmpeg, and a focus on model performance metrics.