Sebastian Hofstetter

Making HyPer, the next generation multi-purpose database engine for Tableau.

Making HyPer, the next generation multi-purpose database engine for Tableau.

Designed and prototyped the event analytics platform for the global information security infrastructure of Allianz SE (Apache Spark, Cloudera-Hadoop, SolR). Key aspects include: - advanced analytics (i.e., anomaly detection via machine learning) - static analytics and alerting for real-time incident management - search indexing for hunting

Ongoing research on secure Cloud Machine Learning via trusted containers in hardware (FPGAs) and software (VSM).

Research on secure Cloud Machine Learning via trusted containers in hardware (FPGAs) and software (VSM). – Designed security and performance study for decomposition of Random Forests into trusted containers. – Improved performance of existing simulator by factor 5 by replacing interpreation with compilation. – The performance cost of security after my optimizations is less than factor 2, using inter- and intra-tree parallelization, cache friendly memory requests, batching and vectorization.

A collaborative consumption start-up that focuses on local trading and peer-to-peer selling, currently associated with about 80,000 users. Allows for trading chains of goods with and without money. – Managed team of 4 developers, set up work flows (Git) and coding conventions. – Designed and implemented backend (Python, App Engine, web2py), created frontend APIs.

Developed a web-scraping solution for non-technical end-users (Python, App Engine) with – automatic form and target data detection – a preprocessor that normalizes internationally varying formats like date, time or currency – support for multithreading, scalability to tasks with 107 and more results – recursive scraping tasks that allow for dynamic crawling of web pages.

Implemented the numerics required to solve the Navier-Stokes Differential Equation for incompressible fluids resulting in a Kinect controlled fluid simulation (C++, Trilinos).

Extended the relational in-memory DBMS HyPer with support for the graph-based query language SPARQL. – Created a SPARQL-SQL crosscompiler and support for graph based input data (RDF). – My implementation on HyPer significantly outperformed its competitors RDF-3X and Virtuoso.

Datenbank Management Systeme, Machine Learning, IT Security

Datenbank Management Systeme, Machine Learning, IT Security