Full-Stack Software Engineer | Java · React · Python | Applied AI, Secure Distributed Systems & Research

Based in Oulu, Finland. I am an M.Sc. student in Software Engineering and Information Systems at the University of Oulu (2026–present), with an earlier M.Sc. in Software Engineering from Azad University, Najafabad Branch. My work brings together full-stack delivery, applied AI, cybersecurity and distributed systems: from Java endpoint agents and React/Python services to reproducible intrusion-detection experiments. I have delivered industry software across security platforms, SaaS and scientific data systems, and my federated-learning work has appeared in The Journal of Supercomputing.

⚙️ Engineering · 🔬 Research · 🏗️ Commercial work · 🔗 Links

I design and deliver React and JavaScript interfaces, Python REST services, data models and deployment workflows. My experience includes modular service design, API/data contracts, integration testing and customer-facing operational systems.

I build bounded, observable automation into product workflows: web crawling, Pydantic validation, resumable processing and optional LLM-assisted resolution where it helps resolve genuinely ambiguous records.

At Fardaad (2022–2024), I helped build a hierarchical patch and vulnerability-management platform: Python services and queues, CVE enrichment and matching, dashboards, and Java agents for Windows and Linux endpoints. The delivery process used GitLab CI/CD; recent public work uses GitHub Actions where present.

My research code covers federated and centralised intrusion detection on Edge-IIoTset, plus an operations-research pipeline for constrained TV-advertising scheduling. These are separate research strands: the latter is optimisation, not machine learning.

• 🧭 Company Domain + Contact Enrichment — a Dockerised Python pipeline that normalises company records, resolves domains with DNS/HTTP checks, crawls sites for contacts and persists phase checkpoints. Pydantic schemas, idempotent phases and a capped optional LLM step make interrupted runs inspectable and resumable. It is a public technical demonstration related to my past 2026 work at Aminyx, not a release of company IP.
• 🌱 BrassicaLNC — a scientific research database containing 1,852 curated lncRNAs from 2,361 RNA-Seq libraries. I contributed React exploration interfaces, Django REST APIs, data models and deployment/operations; the public backend repository documents the code that is available.
• ⚽ Matchline MVP — a public React/TypeScript/Vite product prototype with Zustand and Leaflet. It uses fictional/local data and is not presented as a production sports-data or payment service.

Federated Intrusion Detection is the Flower-based research implementation connected to my M.Sc. thesis and 2026 paper. A server coordinates aggregation of local 1D-CNN updates; experiments cover IID and non-IID client-data configurations on Edge-IIoTset, without centralising raw traffic. See the peer-reviewed paper: Fadaei & Barekatain, The Journal of Supercomputing (2026), DOI 10.1007/s11227-026-08314-x.

Centralised Intrusion Detection is the comparative baseline code, including 1D-CNN and LightGBM experiments for binary, 6-class and 15-class tasks and evaluation measures documented in the repository. It complements the federated work; it is not a separate publication.

This one research collaboration supported a master's thesis project at Sharif University of Technology, Tehran; it was not a Sharif degree, appointment or my own Sharif thesis. The public code is intentionally split into a navigable pipeline:

Synthetic instances  →  Exact optimisation  →  Search alternatives
Data generator       →  MILP / CPLEX         →  NSGA-II · GA · SA · Tabu

1. Synthetic data generator creates constrained scheduling instances: breaks, budgets, reach targets, release/deadline windows and advertiser conflicts.
2. MILP model consumes those instances in Python with DOcplex/IBM CPLEX to seek exact solutions subject to capacity, timing and allocation constraints.
3. Metaheuristics provides NSGA-II, genetic-algorithm, simulated-annealing and tabu-search alternatives.

CPLEX is commercial software, so availability of the repository does not by itself promise solver-independent reproduction.

• CNC Process Management System (2025) — independently built and customer-deployed for a manufacturing workflow; the application is in active production use. Project entry · Closed-source.
• Psychological Clinic CRM (2026) — independently built and customer-deployed for appointments, clinician schedules, patient records, sessions and billing. Project entry · Closed-source; no patient data or screenshots are published here.

These are deployed customer products, not public GitHub repositories. Their links are public entry points supplied by my portfolio/CV; access and any demonstration account remain subject to the product owner’s authorisation.

Publication. Fadaei, A. & Barekatain, B. (2026). “A novel intrusion detection system for dew computing environments based on an enhanced federated deep learning model.” The Journal of Supercomputing, 82, Article 172. DOI.

I have also taught practical programming and software-development projects through Iran TVTO (2024–2026), led backend-development workshops at Dana Cup (2024–2025), and taught an applied Python workshop at Isfahan University of Technology (2021).

Portfolio · LinkedIn · ORCID · Academic CV · Industry CV