Petru Balan - Full Stack Engineer & ML Researcher | Next.js, NestJS, Python
I build high-performance, scalable web applications (Next.js/NestJS) and research Machine Learning models for healthcare. With 1.5+ years of commercial and freelance experience, I specialize in solving business bottlenecks through clean architecture and optimized data layers.
Work Experience
Politehnica University of Bucharest
Machine Learning Researcher
Engineered a custom GRU-D (Gated Recurrent Unit with Decay) neural network to handle missing data in clinical time-series, outperforming baseline mean imputation and reducing error rates by 60%.
Architected a data preprocessing pipeline to clean, normalize, and structure 500k+ raw electronic health records into time-stamped sequences optimized for predictive modeling.
Conducted comprehensive literature reviews on state-of-the-art AI applications in healthcare, establishing the foundational architecture for an ongoing senior thesis focused on early Sepsis prediction.
DIVDOT CODE
Full Stack Developer
Collaborated in a 4-person agile team to build and optimize various SaaS platforms, taking ownership of full-stack feature delivery and real-time system architecture.
Boosted frontend performance by 50% across multiple web apps by migrating to Next.js, implementing Server-Side Rendering (SSR), and eliminating CSS render-blocking resources.
Engineered a cross-platform real-time chat application for web (React.js) and mobile (Flutter), integrating Firebase cloud storage and OAuth 2.0 (Google/Facebook) for secure authentication.
Streamlined CI/CD pipelines by dockerizing application environments and authoring automated deployment scripts, reducing server provisioning time by 30%.