You are viewing content from a past/completed QCon

Presentation: A Cost-Sensitive Approach for Resource Allocation in Virtual Machines

Track: Handling Sequential Data Like an Expert / ML Applied to Operations

Location: Cyril Magnin II

Duration: 1:05pm - 1:55pm

Day of week: Wednesday

Share this on:


Throughout recent years, ING has made a shift from hosting processes on designated physical servers to virtual machines (VM) warehouses. While this transition has contributed to ING’s development teams in providing teams agility and elasticity in resource allocations, the potential for cost reduction on infrastructure spending has not fully been realized. Many VM’s have not been shifting their resource allocation actively according to their usage, resulting in a yearly expense of over 60M EUR on (often idle) computing infrastructure.

In this application talk, Dor will take the audience step by step in the process of building an inner-organizational data science solution. Dor will share insights on the time-series model for predicting usage, the optimization which minimizes costs and risks, the process of deploying data science models to production and some best practices of creating a data science model in an agile methodology.

Speaker: Dor Kedem

Senior Data Scientist @ING Nederland

Dor has over a decade of experience developing big data products for security industries, financial markets and banking industries. His research on metric learning and cost-sensitive learning has earned him publications in NIPS, AISTATS and a monetary prize in Cha-Learn competitions. As a data scientist at ING domestic banking, he is involved with multiple projects modelling consumer and market behavior, optimizing business and IT processes.

Find Dor Kedem at

2019 Tracks

  • ML in Action

    Applied track demonstrating how to train, score, and handle common machine learning use cases, including heavy concentration in the space of security and fraud

  • Deep Learning in Practice

    Deep learning use cases around edge computing, deep learning for search, explainability, fairness, and perception.

  • Handling Sequential Data Like an Expert / ML Applied to Operations

    Discussing the complexities of time (half track) and Machine Learning in the data center (half track). Exploring topics from hyper loglog to predictive auto-scaling in each of two half-day tracks.

    Half-day tracks