Work Experience

Led the development of a Kubernetes-based AI pipeline architecture, integrating model training, inference, monitoring, and secrets management to enhance system efficiency and scalability.
Automated AI-driven CI/CD workflows using Python and Ansible, achieving 95% deployment automation, reducing big data pipeline deployment time to a few seconds.
Designed and optimized a big data processing pipeline supporting AI-based Automated Optical Inspection (AOI) for real-time manufacturing defect detection (~1 million data points/day).
Assisted in optimizing AI model load balancing mechanisms and debugging inference pipelines, improving model stability and system performance.
Managed AI software lifecycle upgrades, ensuring up-to-date frameworks and model deployments with 98% stability confidence for a seamless production environment.
Conducted training sessions on AI model deployment best practices, Kubernetes adoption, and system reliability, including on-site business trips to improve AI infrastructure adoption.

Developed an AI-powered object identification system for small parts (<1cm) using segmentation and filtering in a multi-modal pipeline, enhancing detection accuracy.
Optimized inference time by 30% through model quantization and transitioning image processing from Python to C++ for improved efficiency.
Built a MobileNet-based MVP for rapid prototyping using Streamlit and FastAPI, enabling real-time image analysis and deployment.
Prototyped an iOS-based AI solution using CoreML, UIKit, and ARKit for on-device product identification, leveraging Apple's AI ecosystem.
Implemented ARKit-based measurement techniques for length and thickness-based product identification, enhancing real-world AI applications.
Designed a scalable AWS pipeline with GPU-accelerated microservices, enabling high-performance AI-driven image processing in the cloud.
Explored generative AI techniques by leveraging latent space representations of generative models for image synthesis and augmentation.

Executed ETL (Extract, Transform, Load) processes for energy trading data, resolving type errors and optimizing data conversions for seamless integration.
Automated data ingestion into Azure SQL Database, scheduling it as a cron job to ensure real-time data availability.
Developed an interactive trading analytics dashboard using Dash by Plotly, enabling 30+ traders to perform real-time data analysis and decision-making.
Enhanced data visualization and accessibility, improving trader productivity by 2× through intuitive and dynamic insights.
Conducted regression analysis to uncover key parameter dependencies in trading data, supporting data-driven strategy optimization.

Application developer in PEGA, JAVA built tool
Optimization of the interface between IT operations
Conception and implementation of innovative solutions in the area of enterprise content management