↓Skip to main content
Danial Jafarzadeh Jazi

Curriculum Vitae | Danial Jafarzadeh Jazi

·3 mins·
Author
Danial Jafarzadeh Jazi
AI systems researcher working on GPU-optimized kernels and multimodal transformer architectures, with application to brain state decoding from neuroimaging data (MRI, EEG).
Table of Contents

Curriculum Vitae
#

Download CV as PDF

Danial Jafarzadeh Jazi Tehran, Iran · danialj999@gmail.com · LinkedIn · GitHub

Research Interests
#

AI systems researcher at the intersection of efficient deep learning and computational neuroscience. I develop GPU-optimized kernels and tensor computing frameworks to accelerate multimodal transformer architectures, with application to brain state decoding from neuroimaging data (MRI, EEG). Seeking a research-intensive postgraduate program to advance scalable neural and medical AI systems.

Education
#

B.Sc. in Computer Science — Islamic Azad University, Central Tehran Branch, Tehran, Iran Sep. 2021 – Sep. 2025 (graduated)

  • Cumulative GPA: 17.65/20.00 (approximately 3.5/4.0)
  • Upper-division / Major GPA (last 94 credits): 18.74/20.00 (approximately 3.75/4.0)
  • Relevant coursework: Machine Learning, Artificial Intelligence, Data Structures & Algorithms, Linear Algebra, Probability & Statistics

Research Experience
#

Research Assistant — Islamic Azad University, Tehran, Iran Sep. 2023 – Sep. 2024

  • Multimodal neuroimaging (first author): Proposed a transformer-based framework for brain state decoding from structural MRI data, designing multi-head cross-modal attention mechanisms and positional encoding strategies in PyTorch. Implemented the full training pipeline, with core model components under active development. First-author manuscript describing the proposed architecture available on arXiv.
  • Medical image analysis: Designed and trained CNN-based architectures for brain tumor detection and classification on large-scale MRI datasets, handling DICOM volume ingestion, resampling, normalization, and data augmentation protocols.
  • Mentorship: Supervised and mentored 3 undergraduate researchers on machine learning experimental design, code review, and reproducible research practices within the lab.

Teaching Experience
#

Teaching Assistant — Advanced Programming — Islamic Azad University, Tehran, Iran Supervisor: Dr. Elham Taghizadeh · Jan. 2025 – Sep. 2025

  • Supervised students covering object-oriented design, data structures, and software engineering principles in Python
  • Conducted weekly problem-solving sessions, facilitated code reviews, and assisted in assignment design and grading

Publications
#

[P.1] D. Jafarzadeh Jazi and M. Hajiesmaeili, “Transformers for Multimodal Brain State Decoding,” arXiv preprint, 2025. [arXiv:2512.08462]

See the publications page for details.

Selected Projects
#

GPT Language Model from Scratch — Python, PyTorch, tiktoken · 2025 github.com/Danialjfz/GPT-Language-Model

  • Built a complete GPT-2-style decoder-only transformer (~124M parameters, 12 layers, 12 heads) from scratch in PyTorch with multi-head causal self-attention, learned positional embeddings, and GELU-activated feed-forward blocks
  • Implemented full training pipeline: AdamW with cosine LR schedule and warmup, gradient clipping, mixed-precision (BF16), and periodic checkpointing; achieved validation perplexity ~3.0 on Shakespeare with coherent autoregressive generation via temperature and top-k sampling

GPU Performance Engineering: Opti-GEMM — C++, CUDA, CMake, Nsight Compute · 2025 – Ongoing github.com/Danialjfz/Opti-GEMM

  • Systematic optimization of General Matrix Multiply (GEMM) kernels from naive global-memory implementation through shared-memory tiling, register blocking, and warp-level primitives
  • Empirical cross-architecture benchmarking on Tesla T4 (Turing) and P100 (Pascal), documenting 5× speedup via manual SMEM tiling on cache-less Pascal vs. hardware-cached Turing
  • Profiling with Nsight Compute against cuBLAS baselines

Technical Skills
#

  • Programming: Python, C/C++, CUDA, R, SQL
  • ML & Data Science: PyTorch, TensorFlow, Scikit-learn, Pandas, NumPy, Matplotlib, Seaborn
  • Tools & Platforms: Git/GitHub, LaTeX, Mendeley, Linux
  • Mathematics: Linear Algebra, Probability & Statistics, Optimisation, Operations Research

Language Proficiency
#

  • Persian: Native
  • English: IELTS Academic 7.5 (L: 8.0, R: 8.5, W: 6.5, S: 7.0), August 2025
  • German: A1–A2

Certifications & Memberships
#

  • Elements of AI, University of Helsinki & MinnaLearn — Nov. 2023
  • Writing in Plain Language, LinkedIn Learning — Oct. 2024
  • International Association of Engineers (IAENG), Membership ID: 530744 — Aug. 2025 – Present

References
#

Available upon request.

For academic or research-related correspondence, please contact me.