Senior Product Owner

2/2022 – 1/2025

Programming and Frameworks
• Python Frameworks: TensorFlow, PyTorch, and Hugging Face Transformers for developing and fine-tuning NLP models.
• Data Processing Libraries: Pandas, NumPy, and Dask for efficient data handling and preprocessing.
• Natural Language Processing: NLTK, SpaCy, and SentencePiece for tokenization and linguistic tasks.

Data Preparation and Processing
• Corpus Collection: Aggregating large-scale, high-quality datasets from diverse sources, including therapeutic session transcripts and public corpora (e.g., Wikipedia, OpenSubtitles).
• Preprocessing: Cleaning, normalization, and tokenization of text data, ensuring compatibility with model requirements.
• Validation: Verifying dataset consistency, removing biases, and ensuring linguistic diversity.

Model Development and Training
• Fine-tuning Pretrained Models: Adapting GPT-based architectures for conversational therapy using domain-specific corpora.
• Testing and Validation: Employing test datasets to evaluate model performance, ensuring high accuracy, coherence, and user relevance.
• Optimization: Applying techniques such as quantization and pruning to reduce memory and compute requirements, optimizing performance on limited-resource devices.

Infrastructure and Deployment
• Cloud Infrastructure: Leveraging Azure-hosted solutions with Kubernetes for scalable microservices.
• Resource Management: Designing the system to handle up to 25 concurrent requests per minute with minimal latency.
• Security: End-to-end encryption, GDPR compliance, and robust anonymization protocols for data protection.

Core Development Phases
1. Data Collection and Tokenization: Preparing diverse linguistic datasets and ensuring efficient tokenization for conversational AI.
2. Model Training and Tuning: Iterative fine-tuning with real-world therapeutic scenarios for high adaptability and performance.
3. Testing and Validation: Comprehensive evaluation of conversational flows, edge-case handling, and fail-safe mechanisms.
4. Optimization for Deployment: Ensuring the model’s compatibility with restricted hardware environments while maintaining high throughput.

Teamleiter IT, Full Stack Developer, PMO (IT)

2/2020 – 2/2023

Entwicklung eines Dienstes für Psychotherapeuten, sichere Videokonferenzen, persönliche Konten, Benutzerkonten, API, Datenbank, Microservice-Architektur auf Basis von AWS. Projektteammanagement, Backlog, Stakeholdermanagement.

Webdesign, Apache Kafka, Sicherheit von Webanwendungen, Node.Js, Amazon Web Services (AWS), React (JavaScript library)

5/2016 – 2/2023

Creation of a company, development of unique technical banking products, coordination and integration process. Work with the Central Bank of Lithuania, extensive experience with IBAN, SWIFT, experience in integration with well-known PSPs. Organized the growth of registered users + 17% per month for B2B clients, increased turnover growth by 45% within 1 year, implemented project management (Scrum) and metadalogy (PMBOK). Development and implementation (cloud AWS architecture) of AML, KYC, KYB policies and rules. Development, testing and implementation of reward models for fintech models, gamification experience with rewards. Working with investment companies to raise funding. Transaction support from negotiations with investors to obtaining financing for projects and the company. Company presentations at international exhibitions, for example, at the International Economic Forum in Davos (2019, 2020, 2021)

Webdesign, WordPress, Venture Capital Finanzierung, Sicherheit von Webanwendungen, Java API for XML Processing (JAXP), Jira, Node.Js, Amazon Web Services (AWS), CSS (Cascading Style Sheet), HTML, React Native, Stakeholder-Analyse