Alexandra Bîrzu

Developed a production-ready telephony focused AI Voice Assistant with real time speech processing capabilities: - Built low latency pipeline integrating multiple speech to text and text to speech services. - Implemented interruptible dialogue system using Twilio WebSockets. - Developed multi-tenant architecture handling concurrent voice sessions. - Created modular tool system for triggering different custom actions based on user conversation. Researched solutions for different potential projects: - Researched and experimented with agentic workflows using LangChain. - Researched and implemented web scraping techniques and Retrieval-Augmented Generation (RAG) to enhance a website-based product recommendation chatbot.

Developed and maintained web applications using Laravel, JavaScript, Vue, Inertia, Livewire, and MySQL - Implemented queued jobs to optimize large data preparation and downloads, improving system performance and user experience. - Optimized MySQL queries to enhance database efficiency and reduce latency. - Integrated Pusher to enable real-time data updates and improve user experience. - Utilized Chart.js to create dynamic, real-time, and interactive data visualizations.

Developed and contributed to internal company projects focused on car and license plate detection from images to support a parking management system. - Trained and fine-tuned machine learning models for computer vision and object recognition using TensorFlow and Keras. - Prepared and processed large datasets, applying image augmentation techniques to improve model performance. - Experimented with various neural network architectures to optimize detection accuracy and efficiency.