Dimitrios G.

ESL FACEIT Group is a leading global gaming company operating major competitive esports platforms and tournaments.

• Developed and deployed a user segmentation model using unsupervised clustering to group players by recency and activity frequency, enabling personalized reactivation campaigns, which resulted in 3× higher conversion rates and sustained engagement uplift over 10 days.
• Designed the Fast Skill Calibration model to estimate true skill levels for new players, addressing the Cold Start Problem in matchmaking through supervised predictive modeling.
• Created a Trust Score metric for behavioral analysis over a rolling ten-match window, which helped identify and reward positive player behavior and reduced disruptive incidents.
• Partnered with analytics, product, and growth teams to integrate models into production, supporting decision-making and targeted communications.
• Managed data pipelines in Google Cloud (Vertex AI, BigQuery) for large-scale model training and real-time analytics.

Channel 4 is a UK public-service broadcaster leveraging advanced analytics to optimize audience experience and advertising efficiency.

• Built Impact Forecasting models using supervised learning to predict the performance of commercial breaks by demographic, generating over £8M in additional revenue during 2021.
• Created Ad Load Optimization models to determine ideal ad frequencies per user, maximizing lifetime value while preserving viewer experience.
• Engineered large-scale MCMC simulations using Dask and PySpark, reducing computation times by 40%.
• Delivered explainable models and reporting dashboards for business and sales teams.
• Supported the migration of on-premises data pipelines to AWS Lambda and EC2, streamlining model deployment.

Virgin Media is a UK-based broadband and telecommunications company serving millions of residential and enterprise customers.

• Developed acquisition models using supervised learning to identify high-probability prospects, improving customer acquisition rates by 22%.
• Created proactive churn and upsell models to enhance customer retention and cross-sell effectiveness.
• Designed a feature engineering framework leveraging Dask for parallelized processing on large datasets, accelerating feature computation by 5×.
• Delivered real-time predictive services integrated with marketing automation pipelines.

JustPark is a UK-based smart parking platform connecting drivers with parking spaces through real-time digital solutions.

• Conducted customer segmentation (KYC) using clustering and classification to predict lifetime value and improve user targeting.
• Built demand-supply optimization models that identified profitable expansion regions across the UK, supporting data-driven market growth.
• Produced data reports and dashboards shared with B2B clients to evaluate partnership performance and business viability.
• Automated recurring analytics workflows in Python, saving hours of manual reporting each week.

• Conducted research and modeling as part of MSc dissertation in collaboration with JustPark.

• Designed a customer segmentation pipeline (clustering) using unsupervised learning to identify patterns in user parking behavior.

• Produced actionable recommendations for marketing and retention teams.

• Wrote and presented a dissertation project applying the methodology to real company data, later used internally for campaign targeting.