Faran Taimoor Butt
MS Student · Applied Mathematics & Computer Science · MIPT Moscow

I am an MS student at the Moscow Institute of Physics and Technology (MIPT), where my thesis — GuadPEFT — investigates socially responsible parameter-efficient fine-tuning of large language models, targeting bias amplification and alignment failures during adapter-based adaptation.

Simultaneously, I am a Research Mentee at BASIRA Lab, Imperial College London (I-X) under Prof. Islem Rekik, building GraphFMD — a graph-based benchmark for financial fraud detection targeting NeurIPS. I previously interned at the National Center of Artificial Intelligence (NCAI), contributing a sensor-fusion SLAM system for an autonomous wheelchair.

My research spans machine learning, graph neural networks, computer vision, NLP, robotics, and AI safety. I am actively seeking PhD opportunities for Fall 2026 and welcome collaborations.

Research Interests
Primary
AI Safety & Alignment
Responsible PEFT, bias reduction in fine-tuned LLMs, alignment failures, and value-aligned NLP systems.
Primary
Graph Neural Networks
Graph-based benchmarks, fraud detection, explainable GNN architectures, and biomedical graph learning.
Active
Computer Vision & Medical Imaging
3D volumetric segmentation, real-time inference pipelines, and vision for accessibility and robotics.
Active
Interpretability & Explainable AI
Post-hoc explanations (SHAP, LIME), mechanistic interpretability of transformers, model transparency.
Active
Adversarial Machine Learning
Adversarial robustness, red-teaming of foundation models, and defenses against distribution shift.
Active
Reinforcement Learning
RL for navigation and resource allocation; applications in autonomous systems and intelligent agents.
Applied
ML for Policy & Decision Support
Computational social science; machine learning for high-stakes decision-making under uncertainty.
Applied
Network & Complex Systems
Graph-theoretic modeling of complex networks and social dynamics using statistical ML tools.
Research & Work Experience
BASIRA Lab (I-X) · Imperial College London · Remote
Research Mentee
Oct 2025 – Present
Advisor: Prof. Islem Rekik
  • Selected for a competitive global mentorship in GNNs and explainable AI; advanced to Phase 2.
  • Building GraphFMD — graph-based financial fraud detection benchmark with automated evaluation and public leaderboard, targeting NeurIPS.
  • Conducting research reviews; preparing a full paper from Phase 2 work.
MIPT Moscow · Moscow, Russia
MS Researcher — GuadPEFT (Thesis)
Sep 2024 – Jun 2026
  • Investigating socially responsible PEFT methods for LLMs — reducing bias amplification and alignment failures during adapter-based fine-tuning.
Novosibirsk State University · Russia
Research Fellow
Jun 2025
Advisor: Dr. Evgeniy Pavlovskiy
  • 3D U-Net brain tumour segmentation pipeline on BraTS dataset (Dice = 75%, clinical viability).
  • Deployed real-time FastAPI visualisation platform for segmentation results.
IT Solera · Islamabad, Pakistan (Remote)
Full Stack Development Intern
Jul – Oct 2024
  • Built 5+ AI-powered apps using Next.js, Firebase, OpenAI, LangChain, and Stripe; led 4+ person team to 98% beta accuracy.
  • Mentored by engineers from Amazon, Bloomberg, and Capital One on CI/CD and microservices.
National Center of Artificial Intelligence (NCAI) · Islamabad
Research Intern
Sep 2022 – Oct 2023
Advisor: Dr. Khawaja Fahad Iqbal
  • Autonomous Wheelchair Project: SLAM navigation, stereo vision, and AR-assisted user interface.
  • Sensor-fusion SLAM (laser range finder + stereo vision) — 30% improvement in mapping precision and speed.
Publications
2025
arXiv
Global PIQA: Evaluating Physical Commonsense Reasoning Across 100+ Languages and Cultures
Tyler A. Chang, Catherine Arnett, Abdelrahman Eldesokey, Abdelrahman Sadallah, Abeer Kashar, Faran Taimoor Butt, and 300+ international collaborators.
arXiv preprint arXiv:2510.24081 · Cornell University · 2025
arXiv PDF
2026
In Preparation
GraphFMD: A Graph-based Benchmark for Financial Fraud Detection
Faran Taimoor Butt et al. · BASIRA Lab, Imperial College London
Targeting NeurIPS Benchmarks Track · 2026
Select Projects
GraphFMD
Graph-based financial fraud detection benchmark with automated evaluation and public leaderboard. Built at BASIRA Lab (Imperial College London), targeting NeurIPS benchmarks track.
GNNsPyTorchBenchmarkNeurIPS
GuadPEFT
MS thesis at MIPT. Investigates socially responsible PEFT for LLMs — reducing bias amplification and alignment risks in LoRA-style adapter fine-tuning.
PEFTLLM AlignmentHugging FaceBias
3D U-Net BraTS Segmentation
End-to-end 3D U-Net medical image segmentation on BraTS brain tumour dataset (Dice = 75%, clinical viability). Real-time FastAPI + JS visualisation interface deployed at Novosibirsk State University.
3D U-NetPyTorchFastAPIMedical Imaging
Autonomous Wheelchair — NCAI
Intelligent wheelchair with SLAM navigation, stereo vision, and AR user interface. Custom sensor-fusion SLAM achieved a 30% improvement in mapping precision and speed.
SLAMComputer VisionAugmented RealityROS
AI Agents
Exploration of LangChain and LangGraph agent architectures — multi-agent coordination, tool use, and memory-augmented reasoning for complex task automation.
LangChainLangGraphOpenAIAgents
News
Jan 2026
Advanced to Phase 2 of BASIRA Lab's GNNs for Rising Stars programme. Working on a full NeurIPS-targeted publication on GraphFMD.
Oct 2025
Selected as a Research Mentee at BASIRA Lab, Imperial College London (I-X) for a competitive global mentorship in GNNs and explainable AI.
Jun 2025
Completed a Research Fellowship at Novosibirsk State University. Built a clinical-grade 3D U-Net brain tumour segmentation pipeline (BraTS, Dice = 75%).
Sep 2024
Started MS at MIPT Moscow, funded by the Open Doors Olympiad Scholarship ($20,000).
2025
Co-authored Global PIQA — physical commonsense reasoning across 100+ languages with 300+ collaborators. arXiv:2510.24081.
2023
Won $3,300 across five AI hackathons: Seamless M4T (2nd/265), StableCode (3rd/546), Meta AudioCraft (3rd/293), Llama 2 (5th/1,765), Autonomous Agents (Top-25/2,381).
2022
Represented the Intelligent Robotics Lab at IEEE ICAI 2022, showcasing AI-enabled robotic systems.
Awards, Fellowships & Scholarships
2026
GNNs for Rising Stars Research Training Program
BASIRA Lab · Imperial College London · Selected globally for 8-week GNN mentorship
—
2025
InteRussia Fellowship
Selected among 17 global participants
$1,300
2024
Open Doors Olympiad Scholarship
Full funding for MS in Russia
$20,000
2023
Seamless M4T Hackathon — 2nd Place
265 participants
$1,000
2023
StableCode Hackathon — 3rd Place
546 participants
$1,500
2023
Meta AudioCraft Hackathon — 3rd Place
293 participants
$500
2023
Llama 2 Hackathon with Clarifai — 5th Finalist
1,765 participants
$300
2023
Autonomous Agents Hackathon — Top 25
2,381 participants
—
2022
Kamyab Jawan AI Scholarship
Government-funded AI Diploma · NUST Islamabad
$300
Events & Travels
Your photo at
IEEE ICAI 2022
images/icai2022.jpg
IEEE ICAI 2022
Intelligent Robotics Lab
InteRussia Fellowship
event or ceremony
images/interussia.jpg
InteRussia Fellowship
Selected among 17 globally · 2025
Lab or campus at
Novosibirsk State Univ.
images/novosibirsk.jpg
Novosibirsk State University
Research Fellowship · Jun 2025
NCAI lab or
wheelchair demo
images/ncai.jpg
NCAI Islamabad
Autonomous Wheelchair · 2022–23
Hackathon event
or prize ceremony
images/hackathon.jpg
AI Hackathons · 2023
$3,300 across five competitions
MIPT campus or
Moscow city photo
images/mipt.jpg
MIPT Moscow
MS Programme · Sep 2024 – Present
Certifications
CHFI certificate photo
images/cert_chfi.jpg
CHFI v11 · EC-Council
Computer Hacking Forensic Investigator · 2022
AI Robotics cert photo
images/cert_ai_robotics.jpg
AI Robotics · NCAI, NUST
Artificial Intelligence (Robotics) · 2022
Scholarship / fellowship cert
images/cert_fellowship.jpg
Open Doors / InteRussia
Scholarship & Fellowship certificates · 2024–25
Teaching & Service
Teaching
  • Teaching Assistant · CS163: Computer Programming · Air University · FA2019
Outreach
  • Volunteer AI Educator — AI/ML education to underprivileged school students, Pakistan (2023–2024)
  • Pakistan Red Crescent Society — blood donation, Thalassemia awareness, animal welfare (2019–Present)
Technical Skills
Languages
Python · C/C++ · JavaScript · Java · Swift · CUDA · OpenMP · LaTeX
ML & AI
PyTorch · TensorFlow · Hugging Face · Scikit-Learn · Keras · Optuna · MLFlow · LangChain · LangGraph · OpenAI API · Spark · NLTK
Interpretability
SHAP · LIME · TensorBoard · InterpretML · Captum
Data & Viz
Pandas · NumPy · Matplotlib · Seaborn · Plotly · Power BI · Tableau
Web & Backend
FastAPI · Flask · Express · Node.js · Next.js · React · React Native · UIKit · Firebase
DevOps & Cloud
AWS · Azure · Docker · Kubernetes · GitHub Actions (CI/CD)
Get in Touch

I am always open to discussing research collaborations, PhD opportunities, and ideas across AI safety, graph learning, computer vision, and beyond.

faranbutt789@gmail.com faranbutt@phystech.edu github.com/faranbutt linkedin.com/in/faranbutt
Islamabad, Pakistan · Open to PhD opportunities — Fall 2026