The Sentinel's Dilemma

🧪 Research & Publications

Guarding AI from Hidden Threats - Peer-reviewed research on AI Safety, LLM security, and backdoor attack detection. Dive into the latest discoveries about model robustness, DadmaTools v2, and socially responsible AI systems.

Read Research

Research Focus

📄 CV & Academic Journey

MSc in Computer Engineering at Sharif University BSc from IUST Teaching Assistant for NLP, Artificial Intelligence, Theory of Languages, and Stochastic Processes. Core strength: marrying mathematical rigor with practical innovation.

View CV

Quick Links

🎓 Courses & Learning Path

Curriculum spanning Discrete Math, VHDL, Neural Architecture Search, Persian NLP, and AI Safety alongside targeted Coursera certifications and seminars such as HBC2024.

Explore Courses

Jump to Section:

Research Journey Timeline

2020 (1399) Concours & IUST Admission

Achieved Rank 475 in the National University Entrance Exam (Mathematics & Physics stream) - Top 0.3% among 150,000+ participants. Secured admission to Iran University of Science and Technology (IUST).

📊 475 Rank 🎓 Top 0.3%

2020-2024 Undergraduate at IUST

Full-stack engineering education: Foundations (Math, Physics, Web Design) → Core CS (Algorithms, Advanced Programming) → Systems (VHDL, Hardware Design) → AI Specialization (AI, Computer Vision, NLP). Developed deep knowledge from logic gates to transformer architectures.

🔧 Full-Stack 💻 Engineering

2024 Sharif University Master's

Matriculated into MSc Computer Engineering at Sharif University - the "MIT of Iran." Member of Data Science and Machine Learning group. Selected as Teaching Assistant for Stochastic Processes (Fall 2025), demonstrating mastery of probability theory underlying generative AI and diffusion models.

🎓 MSc Sharif 📚 TA Selected

2024-2025 Bifurcated Research Identity

Domain I - NLP & Mental Health: Contributor to DadmaTools v2 (premier Persian NLP toolkit). Trained Trankit model on Bijankhan corpus. Wav2Vec2 sentiment analysis (94% accuracy). Multimodal sentiment analysis (VGG-16 + LSTM). | Domain II - AI Safety: BackdoorBench analysis, FTT-NAS research, investigating LLM vulnerabilities and defense mechanisms.

🌍 Persian NLP 🛡️ AI Safety

2025 Psychological Health Chatbot & DadmaTools Publication

Published "Psychological Health Chatbot: Detecting and Assisting Patients in their Path to Recovery" - transformer-based conversational agent integrating emotion recognition and DadmaTools infrastructure. Co-authored with Microsoft Research (Sauleh Eetemadi) and University of Ghent (Sadegh Jafari). Published DadmaTools v2 paper at 1st Workshop on NLP for Languages Using Arabic Script.

🧠 Chatbot 📝 Published

Future International Academic Integration

Aspiring for doctoral studies and integration into global Western academic community. International collaborations with Microsoft Research, University of Ghent, and peers. Oxford University identified as target institution. Advancing AI safety research while bridging model security with human-centric AI applications.

🌐 Global 🔬 PhD Ready

Featured Publications & Research

BAIT Paper

🏆 BAIT: Backdoor Detection via Attack Target Inversion

Emerging Scholarship in AI Safety (2023-2024)
Novel approach to detecting hidden backdoors in large language models using attack target inversion. Published research advancing practical defense mechanisms and model robustness.
Impact: Model Security, Adversarial Defense

Read Full Paper

Trustworthy AI

🧠 Psychological Health Chatbot

Affective Computing & Mental Health (2025)
Conversational AI system for mental health support utilizing DadmaTools v2 (Persian NLP), emotion recognition, and transformer-based context management. Co-authored with Microsoft Research and University of Ghent.
Impact: NLP, Computational Linguistics, Social AI

Research Details

Persian NLP

🌍 DadmaTools v2: Persian NLP Infrastructure

Open-Source Contribution to Low-Resource Languages (2025)
Adapter-based NLP toolkit for Persian language. Advanced token classification, dependency parsing, and Named Entity Recognition. Presented at 1st Workshop on NLP for Languages Using Arabic Script.
Impact: Linguistic Diversity, Open Science

View Toolkit

Research Impact & Metrics

3+

Publications

2023–2025

30+

Citations

Google Scholar

2

Major Venues

Conferences & Journals

5

GitHub Repos

Open Source Code

Track detailed metrics and citations on Google Scholar

🎓 View Google Scholar Profile
GitHub Repos

📦 Code & Datasets

Open-source implementations, reproducible research, and curated datasets for backdoor detection and LLM security. Advancing the AI safety community through code.

Explore GitHub

Research Data

📊 Replication Packages

Complete replication materials for published research including code, datasets, and documentation for reproducibility.

Access Data

International Collaborations

🤝 Global Research Network

Partnerships with Sharif University, IUST, Microsoft Research, University of Ghent, and the broader AI safety community. Aspiring for doctoral studies at Oxford and integration into the global Western academic community.

See My Network

Teaching & Mentorship

📚 Teaching Assistant - Stochastic Processes

Fall 2025 TA at Sharif University for advanced stochastic processes. Deep command of probability theory underlying modern generative AI (Diffusion Models, Markov Chains, Brownian Motion).

Schedule Chat

Research Impact

📊 Research Metrics & Impact

Track publications, citations, and scholarly contributions on Google Scholar. Full-stack researcher combining practical software engineering with theoretical depth—from VakilPors legal platform to core AI safety research.

View Scholar

✨ Personal & Beyond

Gallery

📸 Personal Gallery

Photography and personal moments beyond research. A visual journey through travel, hobbies, and life experiences.

View Gallery

Interests

✨ Interests & Hobbies

Beyond academia: photography, travel, reading, and exploring the human side of scientific research.

Learn More