Ali-Asghar Razazpour
Professional Summary
Recent Computer Engineering graduate (February 2026) with strong independent research experience in machine
learning, predictive modeling, and bio-inspired algorithms. Developed novel approaches including a
DNA-inspired
solver for N-Queens (outperforming genetic algorithms for small-to-medium N) and fine-tuned LLMs for
personality
simulation. Passionate about applying ML to real-world domains such as sports analytics and optimization
problems. Seeking Master's admission in AI/ML/Data Science to contribute to innovative research teams
through
RA/TA positions.
Research Interests
- Machine Learning & Predictive Modeling
- Bio-inspired / Nature-Inspired Algorithms for Combinatorial Optimization
- Sports Analytics & Big Data Applications in Sports
- Computer Vision & Multimodal AI Models
Education
B.Sc. in Computer Engineering
University of Bojnurd, Iran
Graduated: February 2026 (9 semesters)
GPA: 15.37/20
Entrance Ranking: 16th out of 60 (Top ~27%)
Relevant Coursework: Machine Learning, Artificial Intelligence, Data Structures & Algorithms, Database
Systems,
Advanced Programming
Research & Project Experience
Fine-tuning Large Language Models for Personality Simulation (Bachelor's Thesis / Senior Project)
University of Bojnurd, 2025–2026
- Fine-tuned transformer-based language models to generate contextually coherent responses mimicking
narcissistic behavioral and linguistic patterns.
- Designed and processed synthetic + real conversational datasets; implemented training pipelines with
hyperparameter optimization.
- Evaluated model qualitatively (coherence, style fidelity) and quantitatively (perplexity, similarity
metrics).
- Technologies: Python, Hugging Face Transformers, PyTorch/TensorFlow, Pandas, Jupyter.
- Outcome: Demonstrated effective personality-aware text generation — foundational work for affective
computing and behavioral AI.
DNA-Inspired Deterministic Algorithm for N-Queens Problem
Independent Research Project, 2024–2025
- Invented a novel bio-inspired algorithm leveraging DNA base-pairing rules to deterministically enumerate
all
valid solutions.
- Achieved complete enumeration of all 92 solutions for N=8; outperformed standard genetic algorithms for
N ≤
20.
- Ongoing optimization for scalability to larger boards.
- Technologies: Python, algorithmic design & optimization.
- GitHub: github.com/Artarazaz/the-question-of-8-Queens
Predictive Modeling of Football Match Outcomes
Independent / Academic Project, 2024–2025
- Engineered end-to-end ML pipeline analyzing 20+ years of European football data for win/draw/loss
prediction.
- Performed feature engineering and trained ensemble models (Random Forest, XGBoost).
- Improved prediction performance over naive baselines via cross-validation and tuning.
- Technologies: Python, Pandas, Scikit-learn, XGBoost, Matplotlib, Jupyter.
Technical Skills
- Programming: Python (Advanced), Java (Intermediate), Git (Advanced), Jupyter
Notebook/Lab (Advanced)
- Machine Learning & Data: ML Algorithms (Strong: Random Forest, XGBoost,
Ensembles), Scikit-learn (Good), Hugging Face Transformers (Good), TensorFlow / PyTorch (Intermediate),
Pandas (Good), NumPy (Intermediate), SQL (Good), Matplotlib (Intermediate)
- Languages: English – Upper-Intermediate (strong technical reading/writing),
German – Beginner
Additional Strengths
- Independent & self-directed research on multiple ML/algorithmic projects
- Creative problem-solving for classic hard problems and real-domain applications
- Technical communication through Jupyter notebooks and academic presentations
- Hands-on experience with real-world and synthetic datasets
- Strong self-management using Git and iterative development