• Time Series AI Development: Engineered unsupervised anomaly detection and forecasting model for multivariate big-data. Implemented sequential architectures such as attention mechanisms and auto-encoders, along with feature correlation analysis to capture cross-feature relationship. • MLOps Infrastructure: Built end-to-end ML pipeline with MLflow for experiment tracking, automate model versioning, and set up production environment for different use case of ML models - real-time and offline… Show more • Time Series AI Development: Engineered unsupervised anomaly detection and forecasting model for multivariate big-data. Implemented sequential architectures such as attention mechanisms and auto-encoders, along with feature correlation analysis to capture cross-feature relationship. • MLOps Infrastructure: Built end-to-end ML pipeline with MLflow for experiment tracking, automate model versioning, and set up production environment for different use case of ML models - real-time and offline depending on the usage.• ML Platform Development: Developed full-stack training platform with React and Flask, enabling real-time model inference and automated hyperparameter tuning via PyTorch; integrated CI/CD pipeline for streamlined deployment. Show less

• Personal Freelance Software Business: https://jastudio-tech.com/• Project Ranges: Developed enterprise-level projects for international clients, including customized ML model training, cross-platform software applications, desktop and mobile full stack applications, and ERP systems.

• Industrial AI Model Training: Reduced missed detection rate by 50% within accepted overkill rate, and fine-tuned models for optimal yield; tested SoTA models, including classification, detection, segmentation. (Project: 3). • AI Software Development: Created frontend and backend of an edge AI Trainer with PyTorch, Flask, React, and Azure, enabling clients to train, finetune, and deploy AI models; Incorporated deployment version control tags (Projects: 4). • Image Storage Platform… Show more • Industrial AI Model Training: Reduced missed detection rate by 50% within accepted overkill rate, and fine-tuned models for optimal yield; tested SoTA models, including classification, detection, segmentation. (Project: 3). • AI Software Development: Created frontend and backend of an edge AI Trainer with PyTorch, Flask, React, and Azure, enabling clients to train, finetune, and deploy AI models; Incorporated deployment version control tags (Projects: 4). • Image Storage Platform Development: Developed frontend and backend of an image storage platform with different user permissions to save, access, preprocess training images with Flask, MongoDB, and React (Projects: 4). Show less

• Government ERP System Development: Built with .NET MVC framework, MSSQL, and AngularJS, implementing hierarchical authority permissions and optimizing performance with SQL stored procedures; CRUD performed on over 100,000+ data entries per query; Deployed the system with MS IIS Web Services (Project: 5).