I teach machines how to listen, understand and respond in various languages.
About me.
I’m a Deep Learning Researcher with the main focus on large scale, multilingual Speech and Natural Language processing.
My past work includes applied Computer Vision research for medical diagnosis, Speech Synthesis, and large scale biometric verification.
I was a recipient of Forbes 30 Under 30 award in Technology field back in 2019.
Background
Staff Researcher
AssemblyAI
11/2021 - Present
I joined AssemblyAI during the Seed Funding stage, as a 6th member of the Research team. In the beginning, I led the efforts to build our Multilingual model suite, end-to-end, from Research to Deployment. My current responsibilities are leadership and on-hands work in Applied Deep Learning Research for large-scale, Multilingual, Semi and Weakly Supervised Speech Processing systems. I have worked on both, Asynchronous and Realtime Speech Recognition networks, as well as on other various Speech Processing and Understanding products. I'm heavily involved in building core blocks of large-scale, multi-billion parameter model training on petabyte scale datasets, using graphic and custom hardware accelerators.
CTO, Head of AI Research
05/2020 - 11/2021
wavetech.
Founded pioneering Speech Synthesis and Recognition startup in Georgia, secured over $400K in investments (equity, grant). My responsibilities were end-to-end building and management of AI Research, Software Engineering and Data Labeling teams and infrastructure, as well as on-hands Research and Engineering. Received heavy media coverage due to the innovative nature of the work. Two highlights that I'm very proud of are: Partnering with and receiving financing from United States Agency for International Development (USAID) and Zinc Network to build world's most accurate Georgian ASR and Diarization system to help them fight Russian State Sponsored fake news and disinformation; Creating ultra-realistic Text to Speech platform and integrating it with largest E-Book store in Georgia - "Saba".
Head of AI Research
09/2020 - 09/2021
CobraID
Led Deep Learning Research for on-edge Facial Detection, Identification and Image Analysis. Built and deployed world's first commercial Masked Face Recognition system that was on par with SOTA Regular Face Recognition research. Optimized and integrated the aforementioned systems in hybrid, edge + cloud computing setting. The system was integrated into Anti-COVID-19 Kiosk, which was deployed in various elite private and governmental organizations, including The President's Administration, Ministry of Justice, Toyota Center Tbilisi.
Research Advisor to Deputy CTO
Liberty Bank
02/2020 - 05/2020
During COVID-19 pandemic, I was recruited by Liberty Bank as an Advisor to Deputy CTO and Technology Executives, to assist them to build Anti-Terrorist and Anti Money Laundering "Know Your Customer" (KYC) systems, as well as to help National Bank of Georgia (NBG, Governmental Financial Regulator) to adapt European-Union drafted KYC regulations. Analyzed various KYC system providers and reported vulnerabilities in their AI-based systems. Negotiated with NBG to increase the quality requirements of AI-based systems. Liberty Bank decided to integrate Spoof Detection and Passport-to-Selfie Face Matching system that I have developed. The software was successfully validated by the National Bank of Georgia, and after integration it helped Liberty Bank to verify hundreds of thousands of users to satisfy Anti-Terrorist and Anti Money Laundering regulations.
Head of AI Research
systemcorp.
02/2017 - 01/2020
Founded and led Georgia's very first Artificial Intelligence Research organization during my student years. Teamed up with the brightest students from San Diego State University Georgia. Had numerous achievements globally and locally, including Microsoft AI Guardian - EMEA Final Gala, Microsoft Imagine Cup Regional Finals 2018 - 3rd place, ~$20,000 Grant from Tbilisi State University for AI Research. Acquired NVIDIA Inception Incubator membership. Managed end-to-end research and deployment of 20+ AI projects. Built world-class Spoof Detection and Face Matching software for financial KYC procedures. Was awarded Forbes 30 Under 30 in Technology field.
Advisor to Vice Chancellor, Research & Sci.
Caucasus International University
11/2018 - 09/2019
Advised to Vice-Chancellor and academic staff to advance the university in research and science direction. Created numerous reports advance their academic research process, concentration and approaches. Developed and integrated software solutions for internal management systems. Analyzed their existing internal and external systems and proposed various opportunities to invest their finances and energy for better future.
Featured On
Media
Television
Conferences
Spotlight Tech Conference 2023
“Do Algorithms Dream Of Electric Sheep?” - Speaker at the largest and most elite tech conference in Georgia to discuss how Artificial Intelligence impacts our lives and what does the future look like.
PyData Yerevan Conference 2022
“Scaling Semi-Supervised Production-Grade ASR on 200 Languages” - Presenting methods and sharing experience of scaling AssemblyAI’s Multi-Lingual systems on large number of languages in the production setting.
Opinion Festival 2023
Speaker on the very first Tbilisi series of worldwide-known festival “Opinion Festival”, to discuss the current state of AI, and how it will shape in the near future
Publications
Contact
Linkedin
in/lchkhetiani
GitHub
@deepconsc
Email
luka@assemblyai.com