Machine learning Engineer

Improved accuracy in architectural plans through segmentation
and regeneration processes, resulting in a substantial reduction of
drawing time by over 3 minutes per plan. Simultaneously, achieved
a 15% enhancement in model accuracy, contributing to heightened
precision in project outcomes.

Implemented an advanced waste collection system leveraging
computer vision, reducing illegal waste in four major Romanian
cities and enhancing recyclability at collection centers.

Deployed and trained 20+ machine learning models for
production, ensuring sustained effectiveness and reliability in
ongoing operations.

Quantized models for edge devices, achieving a notable 25%
speed gain in inference time as part of the “Scaling Intelligence”
initiative. This optimization led to a 5% reduction in operations
costs, demonstrating both efficiency improvements and cost-
effectiveness.

Developed stable diffusion methods for image generation, resulting
in the creation of four distinct architectural plans that effectively
reduced furniture drawing requirements by 4% per plan.

Fine-tuned Large Language Models (LLMs) specifically for
customer service, and innovatively implemented a support system
that autonomously resolved 25% of interactions before requiring
human service intervention.

Created outlier detection and motion monitoring for industrial
machinery, ensuring a smooth workflow. Notably, the system
detected 70% of all problematic pieces of trash that could disrupt
operations, preemptively addressing issues before they reached
the conveyor belt and optimizing overall workflow efficiency