Lamana – PySpark, Python, SQL

The client is a global pharmaceutical distributor. The Lamana project, in which Lamana was involved, aimed to streamline data from logistics, sales, and operations systems into a unified business layer.

• Built an end-to-end Azure-based ETL and analytics platform;
• Designed data flow strategy;
• Implemented bronze-silver-gold data architecture for near-real-time dashboards and forecasting models.
• Mentored junior engineers, and
• Automated ingestion and transformation pipelines that reduced data latency by 60%;
• Improved reporting efficiency across business units.

This project involved a U.S. utility provider aiming to migrate legacy ETL systems to AWS.

• Modernized a large-scale, end-to-end architecture.
• Mentored engineers on cloud best practices.
• Improved data reliability and cut ETL runtime by 50%.
• Developed machine learning models to predict wind turbine faults, enabling proactive maintenance and reducing downtime.
• Achieved a 20% improvement in fault prediction accuracy and shortened maintenance response time.
• Created a chatbot prototype to automate knowledge retrieval and enhance decision-making for engineering teams.

The TV streaming service aimed to modernize its data ecosystem by consolidating fragmented customer and network data. The challenge was integrating real-time data from subscriber activity, network logs, and CRM systems for churn prediction, content personalization, and ad optimization, which was successfully accomplished.

• Defined the data mesh blueprint and coordinated across data domain teams;
• Built distributed data products and domain pipelines in Databricks and AWS, enabling self-serve analytics across marketing, content, and operations;
• Designed ingestion and transformation layers, ensuring interoperability and governance via centralized metadata and cataloging;
• Delivered feature pipelines that improved churn prediction accuracy by 18% and reduced data latency by 40%.

A major retail and FMCG client sought advanced machine learning solutions to improve their sales planning and enhance marketing efficiency across various markets. At the same time, Lamana led a development team to create a marketing budget optimization engine. This engine utilizes predictive models and simulated ROI curves to dynamically reallocate spending across different channels for maximum impact.

• Led a hybrid team of data scientists and engineers, translating business pain points into production-ready ML systems;
• Developed a short-term demand forecasting tool that combined historical sales, promotions, and external signals (weather, seasonality, holidays);
• Delivered actionable insights that drove a 10% improvement in ad ROI and ~$55K annual margin gain;
• Championed AI literacy by training business stakeholders to interpret and apply model outputs in day-to-day decision-making.

An analytics platform powered by AI to assist consumer brands and private equity clients in extracting insights from unstructured online conversations. It analyzes sentiment, identifies emerging topics, and tracks brand perception across millions of posts.

• Led end-to-end development from prototype to production;
• Built natural language processing (NLP) pipelines;
• Improved insight turnaround time by 70% and introduced new data products that increased project revenue by 20%;
• Integrated the analytics into interactive dashboards for consultants to use during client engagements — adding a data-driven layer to brand and market assessments;
• Collaborated with client teams to embed analytics outputs into strategic recommendations.