Open Positions

We are looking for two motivated PostDoc’s to join us in our ERC project FoTran. The positions are open until filled and will cover a period of 2 years each with the possibility of additional extensions thereafter.

In FoTran, we will focus on sentence representation learning and multilingual neural machine translation. The goal is to learn highly abstract language-independent semantic representations from massively parallel data sets and we will test the results on tasks involving natural language inference (NLI) and multilingual (low-resource / zero-shot) machine translation.

Sub-project 1:

  • focus on cross-lingual grounding for NLI
  • develop and evaluate multilingual sequence-to-sequence models
  • vector-based reasoning applied to entailment detection and related tasks

Sub-Project 2:

  • focus on multilingual machine translation
  • transfer learning to support low-resource languages
  • zero-shot learning and a study of vector-based “interlinguas”

Both positions require strong analytic and technical skills with a background (PhD) preferably in computational linguistics or computer science with experience in developing and using models of deep neural networks and sequence-to-sequence models in particular. Previous work on neural machine translation is a plus but not strictly required.

The project will officially start in spring 2018 (probably in March or April) and we would like the PostDoc’s to start as soon as possible after the start of the project. The salary starts at about 41,000 euros gross income per year according to standard rates at Finnish universities and include all social benefits offered to employees of the univeristy. The exact amount depends on work experience and qualifications.

Please, submit your electronic application containing

  • a CV including a list of publications and technical skills
  • a brief research statement with a motivation for the position applied for
  • information about constraints (e.g. practical constraints in relation to starting the job)

Please, send your documents in PDF directly to <jorg.tiedemann AT>. Any questions can be pointed to Jörg Tiedemann using the same e-mail address above.

More information about our team and research is available from the language technology website.