Umar Ashfaq

AI Voice Agent and Machine Learning Engineer with expertise in developing real-time conversational systems and advanced face recognition models using Python and deep learning frameworks. Proficient in speech-to-text and text-to-speech integration, LLM-based response generation, and multimodal data fusion to enhance accuracy and user interaction. Focused on delivering scalable AI solutions that improve system performance and user experience.

Junior AI Engineer at InstantAI Labs (2025-08 – 2026-02)

7-month role delivering client-facing AI solutions

• Developed a real-time AI voice agent using the LiveKit framework and Google Gemini API
• Implemented Speech-to-Text (STT) and Text-to-Speech (TTS) integration for conversational interaction
• Integrated LLM-based response generation for dynamic dialogue handling
• Managed API authentication and quota handling for Gemini API services
• Built and tested the system using Python in a virtual environment
• Developed a face recognition system using ArcFace embeddings to identify individuals from images
• Implemented face detection with OpenCV and extracted 512-dimensional facial embeddings via DeepFace ArcFace
• Applied cosine similarity to compare embeddings for accurate identity matching
• Organized the dataset with labeled folders and stored embeddings for efficient recognition
• Built an interactive Gradio interface for real-time face recognition testing

MS in Data Science – Information Technology University (ITU) (2025-08)

BS in Information Technology – University of Sialkot (USKT) (2020-08 – 2024-07)