Data, Software engineering, and analytics for enterprises across industries.
Design and implementation of a cloud-based data platform on Azure Databricks, migrating on‑premise workflows with a strong focus on cost optimisation and robustness. We build and maintain Python and PySpark pipelines governed with Unity Catalog, and deliver proactive monitoring on defectivity data to reduce escalations.
Migration of Java/Spring based processing to Azure Databricks to modernise the data stack and improve developer productivity. We implemented end‑to‑end pipelines and error analysis dashboards, combining Scala, Python, SQL and Apache Sedona for high‑quality spatial data matching.
Development of large‑scale software and data services that power real‑time location products. We engineered Kafka and Spark based streaming pipelines, containerised services with Docker, and operated them on Kubernetes. We also integrated machine learning models into production flows, handling model serving, inference at scale, and continuous monitoring of model performance. This required bridging real-time data streams with ML inference endpoints, ensuring low-latency predictions while maintaining data quality and model accuracy.
Creation of a modern e‑commerce experience for a jewellery brand, from data‑driven catalogue structure to front‑end design. We focused on clear product presentation, maintainable content, and analytics hooks so performance and customer behaviour can be monitored over time.
Design and implementation of an online presence for a fashion atelier, combining brand‑aligned visuals with a lightweight technical stack. We structured the site and data to make it easy to extend towards online ordering, customer analytics, and future digital services.
Development of a production-grade image classification system on Databricks, handling end-to-end ML lifecycle management. We built comprehensive data pipelines for image ingestion and augmentation, trained deep learning models (CNNs, transfer learning), and registered them in Databricks Unity Catalog with versioning and governance. The platform includes MLflow experiment tracking, automated hyperparameter optimization, and model serving endpoints with A/B testing capabilities. We implemented feature stores for consistent training-serving, batch inference jobs, and continuous retraining workflows on GPU clusters.
