Jan Sedivy

I'm

My vision

Human-like dialogue with artificial intelligence is still elusive. Why? Simple. We’re not there yet. Fifty years after 2001: A Space Odyssey, and forty years after Star Wars, the gap between that level of dialogue and “Alexa - play Spotify” is still a quantum leap. But voice and speech remain the most natural prerequisites for human communication.

My mission

After 18 years in the industry, I have returned to the Czech Technical University, CIIRC, to share my experience with students. My mission encompasses three key areas:


My current research focus is on AI, large language models (LLMs), and conversational AI. In education, I supervise PhD students whose work centers on LLMs, including training specialized, smaller-scale models, reasoning mechanisms, and the design of agentic applications. A significant part of my efforts is dedicated to translating these research outcomes into practical AI-driven applications that enhance productivity in administrative tasks.

Employment, education

I have balanced my professional life between entrepreneurship and academia. Currently, I embrace both roles simultaneously—as a professor at CTU and a co-founder of the startup Promethist.ai.

Professional Experience

Researcher

2010 - Present

Czech Institute of Informatics, Robotics and Cybernetics, Czech Technical University Prague

  • Director of the NLP group
  • Director of the CTU MediLab Foundation,

Technical Lead Manager

2008 - 2010

Google, Switzerland GmBH, Zurich, Switzerland

  • Managing regional teams in Google - Nordics, Central Europe

Research Manager - Voice Technologies and Systems

2000 - 2008

IBM Czech Republic, Prague, Czech Republic

  • Founded, built and led the IBM CZ research and development lab in Prague
  • Research and implementation of the core engine of the Embedded ViaVoice.

Research Staff Member (RSM), Human Language Technologies

1992 - 2000

IBM T. J. Watson Research Center, Yorktown Heights, NY, USA

  • Researcher contributed to key ViaVoice speech recognition algorithms

Recent industry partners projects

2025

Citizen help desk, classification City Hall Prague

2025

RAG base application - Ministry of Finance - CZ

2023

Analytical application - Porsche

2020

Conversational Manual PoC - Skoda Auto

2019

Infotainment visionary PoC - BMW

Amazon Competitions

2025

Amazon Nova AI Challenge - Alquist Coder - second place

2023

Alexa prize - Alquist Social bot - third place

2021

Alexa prize - Alquist Social bot - The WINNER

2020

Alexa prize - Alquist Social bot - third place.

2018

Alexa prize - Alquist Social bot - second place.

2017

Alexa prize - Alexa prize - Alquist Social bot - second place.

Educational experience

Assistant professor

1983 - 1992

Czech Technical University, Prague, Faculty of Electrical Engineering

Education

PhD - Digital Signal Processing

1978 - 1983

Czech Technical University, Prague, Faculty of Electrical Engineering

Master - Electrical Engineering

1972 - 1977

Czech Technical University, Prague, Faculty of Electrical EngineeringY

Startups

Promethist.ai

2019 - present

Promethist.ai - major products Flowstorm.ai and Talk To Poppy

Patents

Below is a list of patents primarily in the fields of Digital Signal Processing, Automatic Speech Recognition, and Internet technologies.

2008

US7,315,613 - Multi-modal messaging

2007

US7,156,309 - Smart book

2006

US7,100,000 - System and methods for processing audio using multiple speech technologies

US Pat. Application 11548976 - Enhancement to Viterbi speech processing algorithm for hybrid speech models that conserves memory

2005

US6,965,773 B2 - Virtual cooperative network formed by local clients in zones without cellular service

US20050143972 - System and methods for acoustic and language modeling for automatic speech recognition with large vocabularies

US20050096070 - Efficient communication with passive devices

US20050086382 - Systems and methods for providing dialog localization in a distributed environment and enabling conversational communication using generalized user gestures

2003

US6584425 - Smart thermometer

2002

US6438247 - Seatbelt Microphone Mounting

US6442519 - Speaker model adaptation via network of similar users

2001

US Pat. Application 9837024 - Systems and methods for providing conversational computing via javaserver pages and javabeans

US Pat. Application 10007084 - Reusable VoiceXML dialog components, sub-dialogs and beans

2000

US6073091 - Apparatus and Method for Forming a Filtered Inflected Language Model for Automatic Speech Recognition

US06023673 - Hierarchical labeler in a speech recognition system

US06016476 - Portable information and transaction processing system and method utilizing biometric authorization and digital certificate security

1998

US05835888 - Statistical language model for inflected languages

1996

US05544277 - Speech coding apparatus and method for generating acoustic feature vector component values by combining values of the same features for multiple time intervals

US05522011 - Speech coding apparatus and method using classification rules

Awards

This is a list of awards, mainly from IBM, recognizing contributions to speech recognition and voice technology development.

2004

The Research Division Award for the development of the Embedded Engine

2000

Outstanding Innovation Award for the development of the Embedded Engine

The Division award for the development of the Embedded Engine

Third Plateau Invention Achievement Award in appreciation and recognition of creative contribution to IBM progress

1999

Second Plateau Invention Achievement Award in appreciation and recognition of creative contribution to IBM progress

1998

Whatever it takes award for management of the speech recognition part of the network car, which IBM exhibited at COMDEX 97

1997

First Plateau Invention Achievement Award in appreciation and recognition of creative contribution to IBM progress

1996

Outstanding Innovation Award in appreciation for the VoiceType 3.0 Design of an algorithm enabling to run CPU expensive labeling on a PC. The current product ViaVoice is still using these algorithms

1995

The Research Division Award for contribution to Large Vocabulary Isolated Speech Recognition

Contact

Location:

CIIRC CVUT 166 36 Praha 6, Dejvice, Jugoslavskych partyzanu 1580/3

Call:

+420 224354181