Amazon hires algorithm specialist from TUT
"Once I observed the growing demand for data mining
expertise among companies such as Amazon.com,
the decision was easy. The principles of data mining
are the same regardless of whether one works in the
public or private sector," he says about plunging into
the world of business.
Timo P. Erkkilä, who has previously worked as a researcher at Tampere University of Technology, joined Amazon this autumn, taking over the development of computational business forecasting tools.
Amazon is one of the world's first online retailers and known as a global forerunner in the development of database systems and new service features.
“My title is forecasting research scientist and part of my job is to develop methods for predicting future sales. My work involves mathematics, programming and data visualization in cooperation with Amazon’s software development teams.”
“Up until now, much of my career has focused on the development of sophisticated algorithms and computational methods for analysing genetic information,” says Erkkilä.
Second time in Seattle
Having worked as a visiting researcher at the Institute for Systems Biology in 2009-2010, Timo P. Erkkilä is no stranger to Seattle.
During his previous visit, Erkkilä began collaborating with a research group headed by Prof. Ilya Shmulevich that explores computational and mathematical approaches for modelling biological systems. The group has been designated as a Genome Data Analysis Center for the Cancer Genome Atlas Project funded by the US government.
“The Cancer Genome Atlas is an ambitious project, aiming to accelerate our understanding of the molecular basis of the most common types of cancer through the application of the latest genome analysis technologies. My role was to develop algorithms for the analysis and interpretation of measurement data."
WHO: Timo P. Erkkilä
- Master of Science in Technology, TUT, 2009.
- Dissertation currently underway.
- At TUT, Erkkilä was part of the Computational Systems Biology research group in the Department of Signal Processing, first as a research assistant and from 2009 to 2012 as a researcher and doctoral student.
- Married, two daughters aged 1 and 4.
- "Family comes first, but I also enjoy pursuing a challenging career and take time to indulge in my hobbies, which include travelling, hiking and photography".
Tools for demanding applications
The work resulted in the development of a suite of tools called Regulome Explorer.
“Regulome Explorer facilitates the interactive visualization of complex data sets through a web browser. Similar exploratory tools are urgently needed, because large-scale undertakings, such as the Cancer Genome Atlas, generate a massive amount of data that is challenging to transfer between research groups.”
In addition, the project spawned a tool known by the acronym RF-ACE.
“I was in charge of developing RF-ACE. The tool uncovers statistically significant associations between multiple variations, such as genetic factors and disease severity, from large and heterogeneous data sets. The associations uncovered by RF-ACE can be used as a resource for Regulome Explorer,” says Erkkilä.
Regulome Explorer and RF-ACE have attracted worldwide interest and were showcased, among others, at the latest Google I/O conference.
Algorithms for multiple uses
Timo says that taking the plunge into the world of business required only minor mental adjustment
“It’s difficult to say whether I’ll return to biological research later in my career. Now my main ambition is to develop algorithms with potential applications across as wide a range of data mining problems as possible. For one thing, I want to advance our understanding of the mechanisms underlying the origin of cancer,” he concludes.