Experienced senior researcher, software engineer, consultant, and lecturer. Led or participated in diverse commercial and scientific research initiatives. Specialized in speech technologies, speech and image processing, HCI, deep learning algorithms, and AI. Leading developer and researcher, and head of the automatic speech recognition (ASR) team. Played a pivotal role in crafting top-regional cloud-based and on-premise ASR solutions including their commercial applications for medical and juridical dictation, and a voice assistant mobile application, alongside the creation of various speech resources. As a chief programmer, pioneered the first high-quality speech synthesizer for Hebrew. Managed multiple projects and project activities within technical development projects. Served as an associate professor and a vice-dean for artistic and scientific research work.
Dedicated machine learning professional, with a proven track record of developing, debugging & testing systems based on cutting-edge AI and ML technologies. Proficient in various IDEs, programming languages, and software tools. Published nearly 100 papers in esteemed journals and conference proceedings, internationally applied technical solutions, and patents. Member of scientific committees for several national and international conferences and a reviewer for international scientific journals. The youngest doctor of technical sciences from the Faculty of Technical Sciences in Novi Sad. Demonstrates exceptional social skills, organizational abilities, and a strong propensity for design. With more than a decade of practical and industrial experience, serves as a regular member of the Centre of Excellence CEVAS, and the leader of the innovation working group of the Serbian AI Society.
Projects:
Project "ELOQUENCE: Multilingual and Cross-Cultural Interactions for Context-Aware, and Bias-Controlled Dialogue Systems for Safety-Critical Applications", Grant agreement ID 101135916, HORIZON-CL4-2023-HUMAN-01-CNECT (January 2024 - Present)
- Project coordinator (UNS)
Project "AI-SPEAK: Multimodal Multilingual Human-Machine Speech Communication", Grant No. 7449, Science Fund of the Republic of Serbia (January 2024 - Present)
- Project co-leader (head of the ASR team)
Project "Innovative Scientific and Artistic Research from the Faculty of Technical Sciences Activity Domain", MESTD No. 451-03-68/2020-14/200156 (January 2020 - December 2023)
Project "MARVEL: Multimodal Extreme Scale Data Analytics for Smart Cities Environments", Grant agreement ID 957337, H2020-EU.2.1.1. (January 2021 - December 2023)
Project "S-ADAPT: Speaker/Style Adaptation for Digital Voice Assistants Based on Image Processing Methods", Grant No. 6524560, Science Fund of the Republic of Serbia (September 2020 - February 2023)
- Project co-leader (head of the ASR team)
Project "SENVIBE: Strengthening Educational Capacities by Building Competences and Cooperation in the Field of Noise and Vibration Engineering", no. 598241-EPP-1-2018-1-RS-EPPKA2-CBHE-JP (November 2018 - November 2022)
- Key staff member
Project "Development of Dialogue Systems for Serbian and Other South Slavic Languages", id TR32035 (January 2011 - December 2019)
- Leader of several project activities (LVCSR, VAD)
Project "DANSPLAT: A Platform for the Applications of Speech Technologies on Smartphones for the Languages of the Danube Region", id Е! 9944 (January 2016 - January 2019)
- Senior researcher
Project "SP2: SCOPES Project on Speech Prosody", SNSF no. IZ73Z0_152495 (April 2014 - March 2016)
Project "S-VERIFY: Advanced Speaker Verification", id Е! 8719 (January 2014 - September 2016)
Project "Central Audio-Library of the University of Novi Sad (CABUNS)", PSNTR No. 114-451-2570/2016-02 (May 2016 - March 2020)
Project "Audio Library for the Disabled (ABOSI)", PSNTR No. 114-451-2210/2011-04 (May 2015 - December 2015)
Responsibilities:
Data Engineering: Managed collection, analysis, processing, labeling, augmentation, quantification, and statistical analysis of diverse datasets (audio/video/text).
Feature Engineering: Applied cross-validation, hyperparameter tuning, regression, dimensionality reduction, feature space embeddings, autoencoders, and CycleGANs.
Model Training and Evaluation: Performed continuous speech recognition, language and acoustic modeling, speaker adaptation, image processing, and emotion recognition using machine learning, deep learning, and data mining techniques; applied various strategies to evaluate model performance.
Optimization: Spearheaded efforts to maximize the efficiency of AI models; achieved significant improvements in processing speed and resource utilization while maintaining robustness and scalability in production environments.
Project Coordination: Project Coordinator for UNS (ELOQUENCE), Head of the ASR Team (S-ADAPT, AI-SPEAK).
Project Management: Oversaw project implementation, organization, and budget.
Dissemination: Contributed to 12 projects and authored around 100 papers in prestigious journals and conference proceedings; developed internationally applied technical solutions and patents.
Lecturing: Delivered lectures on Human-Machine Speech Communication, Selected Chapters in Acoustics and Audio Engineering, Acoustics and Audio Engineering, Acoustics and Audio Engineering in Multimedia, Digital Audio Signal Processing, Optical Telecommunications, Electroacoustics.
Mentorship: Supervised PhD dissertations and master’s theses.
International Collaboration: Worked with international teams of experts to develop and implement advanced AI and ML solutions within complex, high-impact projects.
Professional Affiliations: Regular member of the Centre for Vibro-Acoustic Systems and Signal Processing (CEVAS), group for Acoustics and Speech Technology, accredited as the Centre of Excellence by the National Council for Scientific and Technological Development of the Republic of Serbia on 18 May 2015 and again on 26 February 2020 (October 2014 - Present); member of IEEE Computational Intelligence Society, and IEEE Computer Society (December 2015 - Present); member of scientific committees for national and international conferences, and reviewer for international journals.
Projects:
Project "Digital Audio Signals Processing", AlfaNum - Speech Technologies Ltd (December 2017 - Present)
Project "MEDICTA: Development of Systems for Dictation of Medical Findings in Bosnian/Croatian/Serbian including Latin Expressions", Grant agreement no. 825003, Horizon 2020, DIH-HERO Technology Transfer Experiment Call 2020 (2021 - 2022)
Project "Automatic Speech Recognition System for Dictating Medical Findings", Pension and Disability Insurance Fund of the Republic of Serbia, Contract no. 404.3-399/19 (August 2019 - December 2020)
- Authorized representative
Products and services:
Voice Assistant Application for the Serbian Language, developed by Alfanum - Speech Technologies Ltd (Novi Sad, Serbia)
Automatic Speech Recognition System for Dictating Medical Findings, developed by Alfanum - Speech Technologies Ltd (Novi Sad, Serbia) for the Pension and Disability Insurance Fund of the Republic of Serbia (client)
"100 reasons for 1 click", ASR server for IVR, developed by Alfanum - Speech Technologies Ltd (Novi Sad, Serbia) in cooperation with partners for the Government of the Republic of Serbia (client)
"MEDICTA", A system for dictation of medical findings in Bosnian/Croatian/Serbian including Latin expressions, DIH-HERO Technology Transfer Experiment Call 2020, developed by Alfanum - Speech Technologies Ltd (Novi Sad, Serbia)
"MEDICTA", A system for dictation of medical findings, including stripe mode, dictionary, templates, and personalization options, developed by Alfanum - Speech Technologies Ltd (Novi Sad, Serbia)
"IURISDICTA", A system for juridical dictation, developed by Alfanum - Speech Technologies Ltd (Novi Sad, Serbia)
Responsibilities:
Management: Founded and led a computer programming agency offering consulting, application design, and development services (signal processing, classification, recognition, data analysis, and deep learning).
Solution Architecture: Designed and implemented commercial ASR solutions for medical (Medicta) and legal (Iurisdicta) dictation, a voice assistant application for Android, and various IVR ASR solutions; developed multiple application interfaces and crafted SQL databases.
Stakeholder Interaction: Clarified requirements with stakeholders, provided user education, instructions, and technical support.
Generative AI: Gained expertise in generative AI and prompt engineering for feature development and debugging.
Quality Assurance: Conducted testing, debugging, and code reviews; assessed the naturalness of HCI and customized dictionaries and grammar models according to users’ expectations and feedback.
Safety: Ensured the security and ethical integrity of data, models, and procedures through anonymization, bias mitigation, encryption protocols, and access controls.
Products: LVCSR (Serbian, Bosnian, Croatian, Montenegrin), Medicta, Iurisdicta (in cooperation with AlfaNum)
Technologies: C++, C#, Python, Matlab, Kaldi, PyTorch, TensorFlow, MySQL, TFS, Git, Google Cloud, ChatGPT...
Responsibilities:
Management: Organized scientific lectures and facilitated innovation working group meetings. Serbian Artificial Intelligence Society is a society promoting AI research and the development of applications in the artificial intelligence industry. Members are Serbian AI companies, researchers, decision-makers, entrepreneurs, organizations, professionals, and students active in, or interested in artificial intelligence.
Responsibilities:
Journal Editing: Managed the review process for a special issue entitled "Recent Advances of Computational and Mathematical Applications in Deep Learning"; oversaw the selection and editing of high-quality manuscripts, ensuring adherence to the journal’s standards and relevance to the theme. Axioms is an international, peer-reviewed, open-access journal of mathematics, mathematical logic, and mathematical physics, published monthly online by MDPI.
Responsibilities:
Management: Established conditions for artistic and scientific research activities, overseeing and analyzing outcomes.
Planning: Developed curricula and reports, aligning content with strategic plans, national, and EU higher education standards, and integrating them into the teaching process.
Quality Assurance: Implemented corrective measures and contributed to publishing activities.
Lecturing: Delivered lectures on Audio Engineering, Physical and Physiological Acoustics, Electroacoustics, Applied Acoustics, and Spatial Acoustics with Sound Reinforcement.
Responsibilities:
Lecturing: Delivered lectures on Artificial Intelligence, and Multimedia Information Systems.
Responsibilities:
Data Engineering: Oversaw the collection, analysis, and processing of diverse textual, audio, and user datasets.
Software Development: Designed and developed software for Windows, Linux, and Android platforms, including continuous speech recognition and synthesis, speaker identification, human-computer interaction, and speech segmentation (for Speech Morphing, Inc.).
Quality Assurance: Conducted testing, debugging, and code reviews.
Leadership: Coached, organized, and supervised ASR team members.
Products: ASR (Serbian), Axon Voice Assistant
Technologies: C++, C#, Matlab, Kaldi, Java/Android, JNI, HTML, CSS, Bash, Shell...
Responsibilities:
Lecturing: Engaged in teaching and laboratory practice (Audio Engineering).
Responsibilities:
Software Development: Conducted phrase analysis, input reception, lexicon retrieval, preprocessing, part-of-speech tagging, reading selection, phonetic reconstruction, code optimization, and end-user application development for Aharon TTS.
Safety: Developed data encryption protocols and applications.
Consultancy: Provided consultancy on technical capabilities and application development.
Products: Aharon TTS (Hebrew)
Technologies: C++, C# .NET, Java
Projects:
Project "Development of Dialogue Systems for Serbian and Other South Slavic Languages", id TR32035 (January 2011 - December 2019)
Project "Human-Machine Speech Communication", id TR11001 (October 2009 - December 2010)
Responsibilities:
Research and Development: Conducted research on human-computer interaction, clustering algorithms, digital signal processing, advanced statistics, emotion recognition, speech and image processing, and speech recognition and synthesis for Serbian and Hebrew.
Mentorship: Trained and mentored younger researchers.
Lecturing: Delivered lectures on Automatic Speech Recognition and Synthesis, and Design of Spatial Forms.
Core Network and Services, Network Operations
Responsibilities:
Training: Gained experience with GSM, UMTS, WCDMA, SMS, MMS, and SS7 Protocol; acquired knowledge of wireless network architecture, roaming, and call tracking; assisted with base station repairs and integration.
Reporting: Created technical reports and participated in on-site training.
Computer Engineering Department (RT-RK)
Responsibilities:
Training: Acquired hands-on experience in measuring harmonics at low SNR; gained knowledge of SAADK converters, and DSP techniques.