I help companies build AI systems that drive revenue

I architected the solution and led a team of 5 AI engineers building a multi-agent platform that reads incoming client request emails, searches the CRM and the service portfolio, and drafts a customized proposal reply. I designed the LangGraph orchestration, a parameter-resolver cache that answers repeat requests without extra LLM calls to cut token cost, and a Langfuse-based evaluation and tracing system wired into CI/CD with a visual trace explorer and a feedback-to-eval loop.

I led the architecture and development of an AI platform that reads tender documents, extracts project requirements, matches internal professionals to each opportunity, and auto-builds their CVs.

I planned and built a platform that interprets complex government proposals and scores their socio-economic potential and feasibility, giving decision-makers automated recommendations.

I built ML pipelines in Google Earth Engine to classify satellite data and shipped an interactive app that measures land and water resource risk, with GPT-based interpretation of indicators.

I rolled out an ML accelerator framework across teams to standardize delivery — CI/CD pipelines, Databricks-based development and tracking, and enforced code quality gates.

I built and deployed an MVP that reads thousands of daily sales emails to surface opportunities and automate replies, while managing a small team of analysts and engineers.

I set up the Databricks workspaces and data pipelines, trained forecasting models for visitor volumes, and built monitoring dashboards used by marketing for campaign planning.

I migrated pipelines from Airflow to Azure ML, modularized them into Python packages and Docker environments, and automated retraining for gas station demand and cashflow forecasts.

I owned the full modeling process, from feature engineering to production on Azure Databricks, and presented the retention and investment recommendations directly to C-level executives.

I built the full ML pipeline over large-scale transactional data to score the financial reliability of millions of clients, driving payment terms and credit limits.

I designed forecasting and clustering models over tractor and truck sensor data, deployed them on SageMaker, and integrated the outputs into the fleet management product.