Martin Z.

• ● Flagship Platform (WaterWays-AI): Architected and deployed a comprehensive, enterprise-grade water management SaaS entirely from scratch, handling front-to-back development and cloud infrastructure.

• ● Client-Side Compute Optimization: Successfully adapted and integrated heavy legacy hydraulic simulation engines (EPANET and SWMM) directly into the browser. This eliminated backend compute overhead, drastically reduced server costs, and enabled zero-latency, high-speed simulations for end-users.

• ● IoT Data Engineering: Engineered highly efficient, serverless custom Python ETL pipelines capable of real-time IoT data ingestion, stream processing, and automated anomaly alerting without bottlenecking (20k+ sensors).

• ● Advanced Integrations: Developed dynamic GIS spatial data visualization tools and implemented robust Role-Based Access Control (RBAC) via Firebase auth token.

• ● Performance Optimization & Cost Reduction: Improve legacy imap ingestion pipelines from 40+ minutes per run to 1-2 minutes per run. Implemented aggressive multithreading and parallelization strategies in Python, significantly accelerating data processing speeds and reducing Azure cloud compute costs.

• ● Architecture & Scaling: Designed and deployed machine learning and data engineering pipelines for global-scale IoT devices, utilizing Azure cloud services and managing the Snowflake data warehouse architecture.

• ● Strategic Migration to Microsoft Fabric: Aligned with business stakeholders to resolve severe performance bottlenecks and end-user latency complaints caused by Power BI to Snowflake data imports. Architected and executed a full migration of data ingestion and analytics pipelines to Microsoft Fabric, leveraging Power BI's Direct Lake mode to eliminate continuous data scanning overhead and ensure instant report loading.

• ● Owned the end-to-end lifecycle (from raw data collection to production deployment) of mission-critical Speech-to-Text NLP models for national intelligence.

• ● Built and trained high-accuracy models using TensorFlow and HuggingFace, executing data augmentation and validation to ensure reliability in highly secure environments.