Amazon R&D, Aureliusstraße 2, Aachen, 52064, Germany
+447308130798 • igor@shalyminov.com
Automatic Speech Recognition research
Data-efficient dialogue system bootstrapping
Dialog System Technology Challenge (DSTC) 8 Fast Domain Adaptation, best performing system as per human evaluation
Improving robustness of goal-oriented dialog systems in an unsupervised way
Voice interfaces for applications, dialogue platform • Voice analytics for callcenters, search engine for voice data • Russian National Corpus, search engine for linguistic research
Natural Language Processing course
Query optimization for a term-distributed search engine
Search engine for the Russian National Corpus
N-gram search for the Russian National Corpus
Enhancing the methodology to access the bottom boundary of software pipelined cycle execution time as the application of an algorithm for finding a maximum cycle in a directed graph, a research project in Intel C++ Compiler for Itanium architectures
A software suite for the reliability estimation of the equipment used in nuclear energetics
Thesis : Data-Efficient Methods for Dialogue Systems
Supervisors: Prof Oliver Lemon, Dr Arash Eshghi
Examiners: Dr Andreas Vlachos, Dr Ioannis Konstas
Professional re-training in Data Science and Computer Science
Thesis title: Cell generation module for Prognosis software suite
DSTC8, Multi-Domain Task Completion Challenge Track, Fast Domain Adaptation Task. Best performing system as per human evaluation
A grant towards presenting at EMNLP-IJCNLP 2019 conference
A grant towards presenting at Search-Oriented Conversational AI (EMNLP 2018 workshop) from conference sponsors - Microsoft, Facebook, Google, and TextKernel
As part of Heriot-Watt team, I participated in the development of Alana 2.0, an ensemble-based socialbot augmented with entity linking
As part of Heriot-Watt team, I participated in the development of Alana, an ensemble-based bot for social conversation with neural reponse ranking
An award towards pursuing a PhD degree