Think in systems
Understand the full workflow before optimizing a single component.
I turn complex machine learning challenges into reliable, real-world software systems.
Machine Learning Research · Software Engineering · Applied AI · Computer Vision
I'm Amir, a Machine Learning Research Engineer and PhD candidate with a multidisciplinary foundation in electrical engineering, applied mathematics, statistics, optimization, and software engineering. I develop reproducible, end-to-end AI solutions for complex scientific and real-world problems, from literature review, problem formulation, and data preparation to model development, cross-validation, evaluation, interpretation, and software implementation. My experience spans deep learning, computer vision, generative modelling, Bayesian and graph-based methods, scientific document intelligence, and retrieval-augmented generation. I combine research rigor with hands-on backend engineering experience in Python, JavaScript, Node.js, Django, REST APIs, SQL and NoSQL databases, search systems, asynchronous processing, Docker, Linux, Git, and CI/CD to build reliable, scalable, and maintainable software.
Machine Learning Research✦Software Engineering✦Applied AI✦Computer Vision✦Backend Systems✦RAG & Retrieval✦
01 · About
Engineering with depth, curiosity and context.
Strong software is more than code. It begins with understanding the problem, its constraints and the people who depend on the result.
My path began in electrical engineering, where mathematics, statistics, optimization and embedded systems taught me how to reason about complex systems. Programming turned that foundation into something practical: first through microcontrollers and Python, then through machine learning, computer vision and backend development.
Graduate study strengthened my statistical and research skills, while professional Django development taught me how to build and maintain real software. Today, I work across the full path from literature review and mathematical formulation to model validation, APIs, databases and deployment. I enjoy the difficult middle: turning an ambiguous challenge into a system that is rigorous, useful and built to last.
Make it reproducible
Design processes that can be tested, explained and repeated with confidence.
Protect the evidence
Preserve provenance, validation and context in data-driven systems.
Build for change
Prefer maintainable architecture over short-lived technical shortcuts.
02 · Career journey
One step led naturally to the next.
From telecommunications and embedded control to backend systems, machine-learning research and software engineering—a connected path, not a collection of isolated roles.
2015 - 2019Education
01
Education
A foundation in systems thinking
BEng, Electrical Engineering - Telecommunications
University of Isfahan · Isfahan, Iran
Built a rigorous base in engineering, mathematics, signals and computation, then applied it to Persian-language sentiment analysis for an e-commerce final project.
2017 - 2019Industry
02
Industry
From software to physical systems
Programmer & Electronics Designer
Mobarakeh Steel Company · Isfahan, Iran
Reverse-engineered and designed an electronic control board for rolling-equipment speed regulation, implementing control algorithms and evaluating system reliability.
2019 - 2023Education
03
Education
Machine learning meets real-world deployment
MSc, Electrical Engineering - Telecommunications
Shiraz University of Technology · Shiraz, Iran
Developed deep-learning tools for CT image segmentation and a real-time face-mask detector for Raspberry Pi, then translated model outputs into web-based software for practical use.
2021 - 2022Industry
04
Industry
Building dependable web backends
Backend Developer
Maralhost · Isfahan, Iran
Developed and maintained Django backend functionality, server-side features and data models in a hosting and server-services environment.
2023 - PresentResearch
05
Research
Machine learning for stroke-related brain function and connectivity
PhD Candidate & Research Software Developer
Université de Picardie Jules Verne · Amiens, France
Developing reproducible machine-learning methods to investigate how stroke affects brain function and functional connectivity. I transform stroke-patient and brain-connectivity data into interpretable features, then design and compare statistical, Bayesian, graph-based, and deep-learning approaches for classifying and predicting stroke-related functional deficits. My work covers preprocessing, feature engineering, optimization, cross-validation, visualization, and interpretation, with the strongest methods translated into reusable Python research pipelines.
2023 - 2026Research
06
Research
Advancing applied machine-learning research
Machine Learning Researcher & Author
Computer vision, segmentation and generative modelling
Published and presented work spanning generative face modelling, deep-learning segmentation, image classification and reproducible model evaluation.
2025 - 2026Teaching
07
Teaching
Making technical ideas teachable
Teaching Assistant
University-level Python & machine learning
Delivered approximately 60 hours of instruction, laboratory support and student guidance while communicating technical ideas to multidisciplinary audiences.
2026Projects
08
Projects
Engineering intelligent platforms end to end
DocuMind, PaperMind & Baxter Chess Robot
Independent software engineering projects
Built document-intelligence services, hybrid retrieval and grounded question answering, plus a vision-driven robotic chess system integrating perception, planning and control.
EducationIndustryResearchProjects
03 · Expertise
A broad toolkit, organized around outcomes.
I work across the boundary between research and production, choosing tools for the problem rather than forcing the problem into one stack.
Machine Learning Research
Research-driven modelling with rigorous validation, interpretation and reproducible experimentation.
- Deep learning
- Computer vision
- Statistical learning
- Bayesian methods
- Graph methods
- Optimization
Software Engineering
End-to-end software development across Python and JavaScript ecosystems, from backend services to user-facing applications.
- Python
- JavaScript
- Node.js
- Express
- Django
- Flask
- Next.js
- React
Applied AI & Retrieval
Grounded AI systems that preserve context, source provenance and retrieval quality.
- RAG
- Embeddings
- Vector search
- Hybrid retrieval
- Semantic chunking
- Local inference
Backend, Data & Systems
Reliable backend architecture and practical infrastructure for asynchronous, reproducible workloads.
- REST APIs
- PostgreSQL
- Redis
- OpenSearch
- MongoDB
- Docker
- Nginx
- CI/CD
- Linux
04 · Selected projects
Complex systems, made tangible.
01
Flagship platform
DocuMind
Scientific Document Intelligence & RAG Platform
A self-hosted, multi-service platform for organizing scientific PDFs, retrieving evidence and answering grounded questions with page-level citations.
My contribution
I designed the platform architecture, asynchronous processing, authentication, database integration, document search and local model inference workflow.
- Next.js
- Node.js
- Python
- PostgreSQL
- Redis
- OpenSearch
- Docker
- Ollama
02
Document infrastructure
PaperMind
Scientific PDF Processing Services
Python services that transform scientific publications into structured, machine-readable representations with hierarchy, figures, tables, formulas, references, provenance and quality information.
My contribution
Designed reusable APIs and command-line workflows for validation, indexing, retrieval and RAG pipelines, then integrated them into DocuMind.
- Python
- REST APIs
- Document AI
- RAG
- Validation
03
Robotics & vision
Baxter Chess Robot
Vision-Based Robotic Chess System
A system that detects a human chess move, reconstructs board state, plans a legal response and commands a Baxter robot to move the piece physically.
My contribution
Integrated computer vision, chess-state reasoning, calibration, motion control and hardware interaction into a complete demonstrator.
- Computer vision
- Robotics
- Planning
- Calibration
- Python
05 · Research
Applied research with engineering discipline.
My research connects computer vision, statistical learning, graph-based methods and reproducible scientific computing, with an emphasis on rigorous validation and reliable implementation.
Current doctoral focus
Complex scientific data & robust machine learning
I develop analytical pipelines spanning preprocessing, feature engineering, statistical and graph-based modelling, optimization, cross-validation, visualization and interpretation.
Two international conferencesPhD progress and methodological findings presented to multidisciplinary audiences.
2026arXivVRXU-net: A Deep Learning Approach for Brain Ischemic Stroke Lesion Detection and Segmentation in T1W MRI2025medRxivRAX-NET: Residual Attention Xception Network for Brain Ischemic Stroke Segmentation in T1-Weighted MRI2024medRxivLumbar Spine Degenerative Classification Using YOLO v8 and DeepScoreNet2024arXivEmpowering Medical Imaging with Artificial Intelligence: A Review of ML Approaches for COVID-19 Imaging2023IEEE IJCBSynthetic Face Generation Through Eyes-to-Face Inpainting
06 · Beyond the code
Communication matters, too.
Teaching
Approximately 60 hours of university-level Python and machine-learning instruction, laboratory support and student guidance.
Presenting
Experience explaining methodological developments and research findings to scientific and multidisciplinary audiences.
Languages
Persian/Farsi (native), English (professional proficiency) and French (developing proficiency).
07 · Contact
Let's build something useful.
I'm exploring opportunities as a Machine Learning Research Engineer, Machine Learning Engineer or AI/ML Software Engineer, particularly where scientific rigour, practical engineering and meaningful real-world impact matter.