Describe Data is an InsurTech start-up using alternative data sources, machine learning and AI to understand insurance risks. We have developed a proprietary analytical engine called “Kompreno”. By combining existing client data with a number of external data sources and using template and bespoke algorithms, Kompreno can provide detailed insights into current books of business along with comprehensive risk modelling.

- Assisted the actuarial function of a speciality insurer with their loss-costing of property, casualty and binder business using modern applied statistical techniques such as Bayesian hierarchical modelling- Standardised and aggregated Bordereaux data files to allow insurers a single view of their risk- Advised credit institutions on the use of Survival analysis in credit modelling to aid in IFRS9 compliance.- Provided advice to credit institutions on their data collection and data-readiness for the use of Machine Learning techniques for fraud detection.- Provided short training courses to actuarial departments on the use of Bayesian methods for the claims reserving.

- Working with datasets over the full spectrum of structure and quality to assist clients in realising the value of the data they had collected as part of their normal operations- Helped both internal and external clients to understand the benefits of modern data technologies and the possibilities raised from the development of cutting-edge applied statistical approaches- Worked on Best Estimate Liability valuation assessments for small life insurers as part of their with-profits books of business- Assisted various London Market specialty lines with their pricing and risk costing assessments using techniques such as gradient boosting and other tree methods, Bayesian hierarchical models estimated via HMC posterior sampling and the creation of custom dashboarding to aid in communication of the model outputs.

Assisted the Customer Analytics team on various internal projects to both analyse and improve customer interaction metrics.

- Statistical modelling with an emphasis on the Bayesian version of standard regression techniques- Bayesian modelling of changepoint and process change problems- Survival analysis techniques to help understand internal operational processes- Generation of fake data to assist in understanding modeling approaches- Use and implementation of spatial modeling, geocoding and GIS systems- Investigated the use of open data sources to improve existing models- Implemented a Monte-Carlo approach to estimate the pricing of mortality swaps.- Use of hierarchical modeling to forecast and assess loss reserving across several lines of general insurance business.

- Implemented a trade simulator to understand position evolution and to assist in forecast evaluation- Implemented a backtesting system for the automated NAV calculation of option spread overlay strategies- Automated the generation of PDF reports based on the daily volatility forecasts- Reimplemented much of the forecasting technology for performance and maintainability- Investigated use of ARMA and ARIMA models to estimate forecast error- Implemented an automated system for adjusting volatility forecasts for known future events- Assisted trading with PnL attribution, PnL calculation, and other trade support issues- Implemented short project on high-yield bond portfolio management- Applied existing forecasting models to commodities data- Conducted a study on estimating the long term variance of TSX-60 components and sector ETFs- Automated the population of historical price databases from a variety of no-cost and low-cost sources- Implemented a study on the use of data visualisation to enhance to usability of forecast output- Installed, configured and managed a web-based project management system to assist project development- Converted the whole codebase over to a distributed version control system to allow for multiversion development in the system

- Configured servers, databases and websites, automated tasks through scripting, engaged in basic performance tuning- Maintained and configured servers and developed course completion system, interfacing with the Moodle system

Deployed and maintained hardware and software platform for a CEP system Designed and implemented a fully automated US / Canada Interlisted Arbitrage trading system on a CEP platform (Progress Apama) Developed prototypes for fully automated algorithmic trading strategies based on rebate-trading, pairs trading and mean-reversion Automated systems interface with backend clearing systems

Used GARCH-based models to forecast equity option volatility Used Monte Carlo methods to model the equity option implied volatility skew Developed prototypes to model forecasting errors, risk management systems and dispersion trading

- Taught various basic computer skills courses to business and language students from first to final year- Helped define and shape curricula within guidelines for the institution- Taught quantitative methods courses for both first and second year students- Created and taught a course on innovation management for and MSc in Entrepreneurship