Jobs

Team
This position is part of the Trust and Safety (TnS) business unit. TnS applications proactively prevent fraud, catch fraud, enforce eBay policies, as well as collect & mine data that will help build future Trust and Safety strategies.

Being a large internet company, we obviously have terabytes of data. Our fraud detection systems scan around 100 million transactions in a day. As if that was not enough, the very fraud patterns that we want to recognize - keep changing constantly!

If this sounds like an exciting challenge to you, you may be the perfect person we are looking for.

Responsibilities

1. Understand business problems and define a related learning / analytical problem.
2. Quickly learn eBay applications and data structures.
3. Pull, integrate and transform data from web, application logs and various databases.
4. Define appropriate learning schemes and algorithms to be used.
5. Modify algorithms, or define new algorithms if needed. Working with eBay research labs and other scientists.
6. Conduct experiments, to build models / predictive features.
7. Test, analyze and evaluate results.
8. Train (or retrain) the selected learning algorithms, over appropriate sample.
9. Support models in production by analyzing the results.
10. Work with a cross functional team of scientists and engineers, as well as leading researchers from academia - to further improve detection efficiencies.

Qualifications

1. MS / BS in CS or related engineering field with 2 / 5 years of related experience. Or PhD in science or statistical fields with rich data mining experience.
2. Experience with machine learning techniques such as neural networks, unsupervised clustering, bayesian methods etc.
3. Exposure / strong interest in working with very wide and tall data.
4. Very strong data analysis skills using Perl / Python / SAS or similar technologies.
5. Hands-on SQL and C/C++ or Java experience.
6. Some experience with an ETL tool like Abinitio or Informatica will be nice to have.
7. Basic understanding of statistics.

Contact:
Harshal Deo, email id: hdeo