Sania Nayab

AI Researcher with 3+ years of expertise in LLM reasoning, Knowledge Graph integration, and automated benchmarking. Experienced in Constraint Chain-of-Thought (CCoT) prompting and addressing AI-generated misinformation. Previously at Mediavoice S.r.l., I designed multimodal algorithms to solve industrial AI projects, including AI Enigmistica (Rebus Puzzle generation). Proven ability to drive innovation within cross-cultural teams, bridging the gap between cutting-edge Generative AI research and real-world industrial applications.

Sania Nayab Profile Picture

Professional Experience

Oct 2022 – Present

PhD Researcher

Scuola Superiore Sant'Anna, Pisa, Italy
  • Research Focus: Enhancing LLMs through advanced prompt engineering techniques, integrating KGs for automated benchmark generation, and tackling fake information generation and detection.
  • Thesis: Bridging Reasoning, Knowledge Graphs, and Multimodal Generation. Inferencing Large Language Models Intelligence.
Apr 2025 – Sept 2025

Visiting Researcher

University of Southampton, United Kingdom
  • Research Focus: Conducted a fake information detection study in collaboration with academic partners.
Oct 2024 – Mar 2025

Researcher

Mediavoice S.r.l., Rome, Italy
  • Research Focus: Developed Rebus Puzzle Generation using Generative AI (Multimodal LLMs) for the AI Enigmistica project.
Mar 2021 – Sep 2022

Lecturer

University of Education, Faisalabad-Campus, Pakistan
  • Taught core programming and computer science courses (Web Engineering, Data Structures & Algorithms, ERP, Introduction to Programming, and CS Fundamentals).
  • Developed course content and assessments aligning with Outcome-Based Education (OBE) standards.
  • Managed and evaluated students' projects and assignments, fostering practical skills and deep understanding.
May 2015 – Mar 2021

Lecturer

University of Faisalabad, Pakistan
  • Facilitated learning in Programming Fundamentals, Database Management, Web Design & Development, Artificial Intelligence, Visual Programming, and Data Structures.
  • Evaluated students' performance and guided them in programming projects, developing their practical skills.
Nov 2016 – Mar 2021

Department Coordinator

University of Faisalabad, Pakistan
  • Managed departmental documentation for QEC and NCEAC HEC Pakistan accreditation processes.
  • Coordinated seminars, workshops, and departmental events. Served as IT Society Head.
  • Acted as lead facilitator for department planning and represented the Computational Sciences department during orientation and exam sessions.

Education

Oct 2022 – Present

PhD in Emerging Digital Technologies

Scuola Superiore Sant'Anna, Pisa, Italy
  • Thesis: Bridging Reasoning, Knowledge Graphs, and Multimodal Generation. Inferencing Large Language Models Intelligence.
  • Field of study: Large Language Models (LLMs), Generative AI.
  • Research focus on advanced prompt engineering, KG-based automated benchmark generation, and fake information generation and detection.
Sep 2013 – Dec 2015

Master of Science in Computer Science (MS)

University of Agriculture, Faisalabad, Pakistan
  • Thesis: Comparative Study of Service-Oriented Requirement Engineering (SORE) Features for Service-Oriented Architecture.
  • Developed a modified SORE model and tested its efficacy within a banking organization.
Sep 2010 – May 2013

Master of Science in Computer Science (M.Sc.)

University of Agriculture, Faisalabad, Pakistan
  • Project: Wellcare Hospital Management Information System.
  • Developed a hospital management system handling patient records, doctors' appointments, and equipment records.

Research Publications

Leveraging Knowledge Graphs and LLMs for Structured Generation of Misinformation
Nayab, S., Simoni, M., Rossolini, G.
Availability, Reliability and Security. ARES 2025. Lecture Notes in Computer Science, vol. 15997, Springer, Cham (2025).
DOI
KGQuest: Template-Driven QA Generation from Knowledge Graphs with LLM-Based Refinement
Nayab, S., Simoni, M., Rossolini, G., and Saracino, A.
KES-IDT 2026 Proceedings of KES Smart Innovation Systems & Technologies Series Springer (2026).
Link
Concise Thoughts: Impact of Output Length on LLM Reasoning and Cost
Nayab, S., Rossolini, G., Simoni, M., Saracino, A., Buttazzo, G., Manes, N., and Giacomelli, F.
Link
Aspect-Context Level Information Extraction via Transformer-based Interactive Attention Mechanism for Sentiment Classification
Nayab, S., Hanif, M. K., Talib, R., and Sarwar, M. U.
IEEE Access, vol. 11, pp. 57683-57692 (2023).
DOI
HarX: Real-time Harassment Detection Tool using Machine Learning
Rizwan, K., Babar, S., Nayab, S., and Hanif, M. K.
In 2021 International Conference of Modern Trends in Information and Communication Technology Industry (MTICTI) (pp. 1-6), IEEE (2021).
DOI

Research Focus

Constraint Chain-of-Thought (CCoT)
Prompt Engineering
Exploring prompt-engineering techniques that aim to refine reasoning and structured prompts to improve LLM intelligence and enhance reasoning precision.
KG-Driven Benchmark Generation
Automated Benchmarking
Leveraging Knowledge Graphs to produce template-driven, verifiable QA benchmarks that assess LLM as judge at scale. This approach is central to the KGQuest initiative.
Structured Misinformation
Generative AI Safety
Investigating the potential of LLMs combined with Knowledge Graphs to both generate and detect structured misinformation, contributing to the field of generative AI safety.
Multimodal Puzzle Generation
Industrial R&D
Applying multimodal LLMs to the generation of Rebus puzzles as part of the AI Enigmistica project at Mediavoice s.r.l., demonstrating the intersection of creative industry challenges and generative AI research.

Courses and Certificates

Building Agentic RAG with Llamaindex
DeepLearning.AI
Completed June 2025
Knowledge Graphs for RAG
DeepLearning.AI (Coursera)
Completed Oct 2024
Generative AI with Large Language Models
AWS & DeepLearning.AI
Completed Nov 2023
Natural Language Processing with Classification and Vector Spaces
Coursera
Completed Sept 2020
Python 101 for Data Science
Cognitive Class.ai
Completed April 2020
Python A-Z: Python for Data Science with Real Exercises
Udemy
Completed April 2020
Big Data 101
Cognitive Class.ai
Completed March 2020
Text Analytics 101
Cognitive Class.ai
Completed May 2020

Conferences and Seminars

ARES 2025: BASS 5th International Workshop on Behavioral Authentication for System Security

20th International Conference on Availability, Reliability and Security, Ghent, Belgium (11-14 August 2025)

MTICTI 2021: 1st International Conference on Modern Trends of ICT Industry

Alrazi University, Sana'a, Yemen (04-06 December 2021)

14th IEEE ICOSST 2020

International Conference on Open Source Systems and Technologies, Pakistan (16-17 December 2020)
Academic seminar
Seminar Session
Academic seminar
ARES 2025 · Ghent, Belgium

ProQuest's Introduction to Publishing in Academic Journals – Part 2

ProQuest Education in partnership with HEC Pakistan (15 December 2020)

ProQuest's Introduction to Publishing in Academic Journals – Part 1

ProQuest Education in partnership with HEC Pakistan (08 December 2020)

HEC Digital Library Collections for Scholarly Research

ProQuest Education, HEC Pakistan (14 October 2020)

Machine Learning for Signal Processing and Big Data Analytics

IEEE Computer Society, FAST National University of Computer and Emerging Sciences (08-09 October 2016)