Dubai, United Arab Emirates
I am a Machine Learning Researcher passionate about applying AI to solve real-world challenges. With a strong foundation in machine learning, deep learning, and data analytics, I have developed impactful solutions across audio, vision, and language domains—such as bird call classification using OpenAI’s Whisper model, AI-powered plant disease detection, and a domain-specific Retrieval-Augmented Generation (RAG) chatbot using LLaMA 3. I recently completed my MSc in Machine Learning from MBZUAI (CGPA: 3.95/4.0), where I focused on advanced topics like open-vocabulary object detection, parameter-efficient fine-tuning of LLMs, and semi-supervised learning for waste recognition. My research has led to publications and submissions at top conferences including INTERSPEECH and BMVC. Proficient in Python, SQL, Azure, and PyTorch, I bring a blend of academic rigor and practical implementation to AI systems, aiming to contribute to impactful and scalable solutions.
• Own lead management and sales funnel data products for 10+ automotive brands, translating KPI and reporting requirements into scalable backend data models. • Build and maintain gold-layer data products in Databricks using SQL and Python, with ownership of metadata, data quality checks, and backend enhancements. • Collaborate with business stakeholders and BI developers to deliver standardized Power BI dashboards aligned with backend data logic and reporting needs. • Lead UAT and deployment of reporting enhancements, while modernizing legacy tables and fragmented reports into unified data products and standardized dashboards.
• Built a representative bird/no-bird benchmark dataset from ~500 hours of field audio by segmenting recordings, using signal-processing–based activity detection for targeted sampling, and manually annotating 500 clips. • Evaluated and compared pretrained bird audio classifiers using precision, recall, and F1, selecting the best performing model for real-world deployment.
• Benchmarking and fine-tuning multilingual LLMs (LLaMA, Mistral, Qwen, GPT-4o) on Gulf-specific datasets, improving factual grounding, bilingual reasoning, and domain adaptation for climate intelligence. • Building and integrating domain-adapted models into an agentic system with modular toolchains and backend APIs, enabling real-time climate knowledge.
As an active member of the Graduate Student Council, I collaborate with fellow council members to support the graduate student community. Our responsibilities include organizing social events that foster networking and engagement, gathering feedback to address student concerns, and taking actionable steps to improve the overall student experience.
As a Teaching Assistant for the 'Foundations of Artificial Intelligence' course for master’s students, I lead weekly lab sessions, facilitate hands-on learning, and provide guidance on fundamental AI concepts. My responsibilities include grading assignments and exams, ensuring fair assessment and timely feedback. Additionally, I mentored student groups on their course projects, helping them apply AI principles to real-world challenges.
• Developed and deployed a real-time face recognition system using Azure Face API and OpenCV, enabling live video frame annotation and identity tagging with high accuracy. • Built and integrated a Retrieval-Augmented Generation (RAG) system into a web app using GPT-4.0 and Azure OpenAI Services, enabling precise, domain-specific question answering over custom documents. • Gained practical expertise across the Azure ecosystem, including AI services, networking, and security; earned Microsoft Certified: Azure Fundamentals (AZ-900).