M Adnan

Responsibilities: • Collaborated with teams of product managers, designers, and developers to define and meet project specifications • Developed and maintained robust Python back-end code for various applications, ensuring scalability and performance • Designed, trained, and optimized DL models, LLMs and NLP applications, pushing the boundaries of AI to enhance user engagement and operational efficiency • Designed, trained, and optimized DL models, LLMs, and NLP applications to enhance user engagement and operational efficiency • Led R&D on new AI innovation for improving problem-solving capabilities and business outcomes • Debugging code, bug fixing, clearance from QA, documentation and third-party evaluations Achievements: • Developed a Data Leak Prevention application that integrates OCR and utilise advance ML and Transformer-based algorithms to automate the detection of data breaches with high precision • Created an advanced semantic search application using LLM-based embeddings and Elasticsearch as a vector database • Implemented ETL pipelines with Elasticsearch to optimize data storage, search, and real-time analytics • Fine-tuned an LLM for a question-answer system, for internal decision-making by providing accurate answers to company-specific queries • Built and deployed RAG application on 80M docs, enhancing retrieval processes using LLama-2 LLM and Elasticsearch • Developed a web application integrating a 70-billion-parameter LLM in offline mode • Designed an automated data summarization tool by fine-tuning multiple LLMs • Developed an image classification application for validation of employee-submitted passport photos, ensuring adherence to company standards • Created a sentiment analysis application to analyze employee sentiments from extensive private company data • Perform real-time ship detection and classification in low visibility • Engineered multiple Python applications and APIs, dashboard creation, leveraging Elasticsearch and PostgreSQL

Working as a freelance Machine Learning Engineer, delivering end-to-end AI solutions across diverse domains including NLP, computer vision, and multimodal systems. • Developed a multimodal content authentication system to verify the authenticity of text, audio, video, and image data using advanced AI techniques. • Built an opinion search chatbot leveraging social media data to analyze and extract public sentiment about products and brands. • Designed and implemented end-to-end prompt engineering pipelines for a fitness application, enabling personalized diet and workout recommendations. • Developed RAG-based applications for complex domains, including Hebrew financial documents and enterprise knowledge systems. • Built an advanced Report Generation System (RGS) using RAG to automate structured document creation by synthesizing information from multiple sources. • Created a Gradio-based application for multi-label protein classification in medical imaging workflows. Engineered a real-time Automatic Number Plate Recognition (ANPR) system with database integration. • Fine-tuned LLMs for customer support automation, improving response accuracy and user experience. Implemented RAG pipelines using LLaMA-2 and Elasticsearch for scalable semantic search solutions. • Developed domain-specific summarization systems by fine-tuning multiple transformer models. • Built sentiment analysis systems for large-scale private datasets, enabling actionable business insights.

Scored 3rd Position.

Pakistan Engineering Council (PEC) # 5834 https://verification.pec.org.pk/v/eV/sED/i.aspx?eid=31312F303035383334