Asadbek Iskandarov

Designed and implemented end-to-end AI pipelines that convert unstructured inputs like PDFs, spreadsheets, and audio into structured actionable data. Developed logic for product identification, attribute extraction, and dynamic cost evaluation using LLM-based approaches. Delivered a responsive AI-assisted frontend interface enabling real-time feedback, structured editing, and human-in-the-loop control to create an intuitive user experience blending AI output with manual refinement.

Automated CRM and data entry operations by integrating Google Sheets, Apps Script, and HubSpot API, synchronizing over 1,000 data points in real-time and saving 20+ hours per week with 100% accuracy. Built and fine-tuned LLM-driven candidate ranking models (LLAMA2, MiniLM) on a dataset of 30,000+ CVs, improving match accuracy by 40% through score-based recommendation pipelines.

Collaborated with engineers and researchers to develop and deploy machine learning and computer vision algorithms for autonomous drone navigation using deep learning (CNNs) and reinforcement learning for real-time decision-making and collision avoidance. Integrated AI models into embedded drone systems, applying Python, TensorFlow, and OpenCV for flight path optimization, sensor fusion, and automation in dynamic environments.