Demajh
Brian Demajh, AI/ML Engineer

“We pioneer R&D breakthroughs in AI/ML to solve impossible challenges in high-stakes domains.”

Innovations unlocked. R&D accelerated. Missions accomplished.

Demajh is an R&D-focused AI/ML firm specializing in innovative, scalable solutions for high-stakes environments—including government and defense applications alongside scrappy startups and Fortune-100 enterprises. From cutting-edge prototypes to robust production systems, we diagnose, develop, and deploy mission-critical AI technologies at break-neck speed.

R&D Capabilities

Reinforcement Learning

Developing advanced RL algorithms for autonomous agents in complex, dynamic environments, including policy gradients, actor-critic methods, and curriculum learning for high-stakes control systems.

MLOps & Infrastructure

Designing CI/CD pipelines, GPU orchestration, and compliance-grade deployments that scale from prototypes to global production clusters.

Multi-Sensor Fusion

Integrating lidar, vision, and IMU data with Bayesian filtering and trajectory optimization for robust perception and decision-making in uncertain settings.

Computer Vision

Building CV pipelines for process mining, fraud detection, and identity verification, leveraging mixture-of-experts models and synthetic data generation.

R&D Case Studies

Vision-Based Process Mining

  • Deployed CV pipeline tracking operator-machine interactions across four factories
  • Reduced safety incidents by 43% and saved $2.1M/yr
  • Guided $100M US reshoring effort from China

Real-Time Churn Prediction

  • Unified 7 SaaS streams into Snowflake lake-house with XGBoost + SHAP
  • Production system for 150+ mid-market customers
  • Supports Series A company with explainable churn/renewal predictions

Identity Verification Fraud Detection

  • Built MoE CV models outperforming AWS Rekognition on face-match tasks
  • Developed synthetic data generator for face-swap detection on IDs
  • Checks >50K jobs daily for fraud in production

Team

Brian Demajh

Brian Demajh

Founder & Principal AI/ML Engineer

PhD in Computer Science from UC Santa Barbara (Deep Reinforcement Learning for Control of Neural Systems). 10+ years in ML, with expertise in PyTorch, AWS, Snowflake, and others. Led R&D at Madrona Venture Labs and Surudo, delivering seed-funded products and VC-backed pilots.

LinkedIn →

Latest Posts

AI Coders Won’t Save You

Value Generation By Data Science in the Era of AI →

Read →

vRAG

Visual Retrieval‑Augmented Generation →

Read →
View All Posts →

Ready to Accelerate Your R&D?