Skip to content
Back to Projects
ML / AI 2024-06

AI-Driven Image-to-Text CI Automation

AI-powered image-to-text CI pipeline using Hugging Face Transformers for automated visual validation, reducing downstream defects by 15%.

15% Fewer Defects
Hugging FaceComputer VisionCI/CDGitHub ActionsPython

Overview

Piloted an AI-driven image-to-text CI automation system at CVS Health using Hugging Face Transformers. The pipeline automatically validates visual UI components during CI builds, catching visual regressions before they reach production.

Key Features

  • Automated Visual Validation: Hugging Face vision-language models analyze UI screenshots during CI
  • Production-Grade Pipeline: Established enterprise-ready pipelines for organization-wide rollout
  • GitHub Actions Integration: Seamless integration into existing CI/CD workflows
  • Defect Prevention: Catches visual regressions that traditional snapshot testing misses

Impact

  • 15% reduction in downstream defects
  • Shifted visual validation left in the development pipeline
  • Enabled automated QA for UI-heavy components at enterprise scale

Tech Stack

  • AI/ML: Hugging Face Transformers, vision-language models
  • CI/CD: GitHub Actions, custom pipeline orchestration
  • Backend: Python
  • Testing: Automated visual regression detection

Related Projects

ML / AI

RAG Support Assistant

Internal RAG-powered support assistant with Slack integration, OpenAI, and ChromaDB. Reduced manual ticket resolution time by 20% at CVS Health.
ML / AI

MRI Brain Tumor Detection & Segmentation

Multimodal MRI classification and segmentation model trained on BraTS dataset with shared encoder, achieving 91.3% accuracy and 97.1% sensitivity.