Langat Vincent

Engineered and deployed an active-learning-based Isolation Forest model, integrating domain-specific client feedback to iteratively improve accuracy and reduce false positives.

Co-developed DIVA and LEAP, flagship fraud and anomaly detection platforms for capital markets, building scalable, adaptive models that significantly improved detection accuracy and reduced false positives.

Enhanced LEAP application's backend and frontend capabilities through optimized API design and system architecture (Python, Django, Django REST Framework), improving platform performance, scalability, and developer efficiency.

Designed and implemented an explainability framework (SHAP, LIME, gradient-based methods) within the LEAP platform, providing clear, localized insights into anomaly detection decisions, boosting stakeholder trust and ensuring regulatory alignment.

Collaborated with stakeholders to define and prioritize product requirements, ensuring alignment with business objectives and maximizing product impact through targeted feature delivery.

Conducted research and development of advanced anomaly detection techniques for financial and credit data, integrating explainability frameworks to enhance model transparency and interpretability.

Developed and deployed robust credit scoring (XGBoost) and fraud detection (CatBoost) models leveraging TransUnion and SEON data, significantly enhancing loan approval accuracy and strengthening fraud prevention measures.

Designed and implemented an automated loan default dashboard, providing daily, weekly, and monthly tracking of default rates, enabling proactive risk monitoring and informing continuous improvements in credit risk models.

Automated payment and debt collection reporting for back-office teams using PostgreSQL, streamlining operations and improving collection outcomes while reducing processing time.

Leveraged in-depth SQL querying (DBeaver, PostgreSQL) for data extraction and analysis, integrating insights into Preset for automated dashboards to enhance reporting efficiency and enable real-time, data-driven decision-making.

Led and mentored a team of junior analysts, fostering skill development and accelerating the successful delivery of data-driven solutions in business intelligence, fraud detection, and credit risk analysis.

Collaborated cross-functionally to align technical solutions with business objectives, integrating fraud detection, credit risk, and compliance strategies for cohesive, high-impact outcomes.

Led data analysis and reporting efforts for two climate-focused pilot projects (fibre production from pineapple waste; carbon credits from biochar production), supporting data-driven decision-making and donor compliance.

Authored comprehensive technical documentation using statistical methods and data visualization, ensuring adherence to scientific reporting standards and data accuracy.

Compiled qualitative and quantitative data to effectively demonstrate project impact, developing compelling case studies and success stories highlighting socio-economic benefits.

Ensured timely and accurate grant reporting in compliance with SMEP/DEFRA requirements, coordinating data gathering with project stakeholders.

Implemented robust data validation processes, upholding data integrity and accuracy across all reporting deliverables.

Conducted energy consumption analysis to identify efficiency opportunities and support sustainability-focused decision-making.

Identified and implemented process improvements, automating and streamlining data workflows, and integrating best practices in data analysis and storytelling.

Designed and implemented a comprehensive network of low-cost air quality sensors across Nicosia, enabling spatial analysis and accurate source apportionment of air pollution to improve understanding of pollution dynamics.

Developed and implemented automated data pipelines for real-time data collection, cleaning, and analysis, ensuring continuous monitoring of air quality and enabling data availability for public health decision-making.

Developed custom tools and dashboards for advanced real-time air quality data visualization, providing stakeholders with dynamic monitoring and crucial insights for public health responses.

Developed and deployed advanced ML calibration techniques for low-cost air quality sensors, significantly improving measurement accuracy and reliability of pollutant data in urban environments.

Performed spatial and temporal analyses of air pollution in Nicosia, identifying key hotspots and patterns to inform environmental policy-making and urban planning in high-risk areas.

Established data quality indicators aligned with international air quality monitoring standards (US EPA, EU Directive 2008/50/EC), ensuring regulatory compliance and contributing to global best practices.

Analyzed sales, customer feedback, and production data to identify trends and optimize marketing strategies.

Developed AI models to improve tea production efficiency and forecast sales, enhancing inventory management and reducing production costs.

Delivered actionable insights and reports to stakeholders, driving informed decision-making and business growth.