Noman Asif

• Architected end-to-end RAG pipelines processing 350+ documents per customer with automated ingestion, re-ranking, and document-level role-based access control.
• Built Neo4j Knowledge Graphs with 10, 000+ nodes and relationships per customer from live and historic data, with semantic query matching powering LLM-driven SQL and Cypher generation for insights extraction.
• Trained YOLOv12 on DocLayNet for intelligent document extraction (tables, images etc from PDFs) with LLM-based conversion to structured markdown.
• Reduced Knowledge Graph construction time from 3 days to under 4 hours for 50, 000+ records through multiprocessing and multithreading, and improved chatbot latency from minutes to seconds via async conversion and multi-layer caching.
• Developed FastAPI endpoints and Streamlit interfaces deployed on Azure Function Apps, serving 50+ active customers. Managed LLM fine-tuning, content filters, and model operations on Azure OpenAI.
• Integrated ElevenLabs (TTS/STT) into the chatbot to enable voice based interaction, expanding accessibility for enterprise customers.
• Engineered hierarchical RBAC with parent-child customer management across Neo4j, PostgreSQL, and MySQL, alongside OTP and MSAL authentication.
• Led and mentored a team of 6-8 AI engineers, conducted technical interviews, and onboarded multiple engineering teams.

• Conducted and graded quizzes, assignments, and course project demonstrations for parallel programming using OpenMP, MPI, and C++.
• Provided guidance to students on parallel algorithm design, debugging, and performance optimization.