Big data & data science
Architecture of large-scale data platforms and analytics, from batch and streaming ingestion to machine learning in production.
Open to new roles · Milan or remote in Europe
Hello, I'm Vahid.
Seyed Vahid Ghayoomie
Computational Systems Engineer
Big data · Cloud, DevOps & MLOps · Security & reliability · Machine learning · HPC · Bio-inspired AI
More than 15 years of experience designing, building and operating large-scale systems, from banking data platforms and security to cloud infrastructure, machine learning and high-performance scientific software.
Politecnico di Milano
- 15+years in software and IT
- ~5×lower cost for a national bank's data platform
- 7years in application security
- 20%of every week goes to research
About me
Engineering across data, cloud, security and AI
Computational systems engineer with interdisciplinary expertise in big data architecture, cloud infrastructure, DevOps and MLOps, security and reliability, data science, machine learning, performance-oriented software development, and bio-inspired AI.
Over more than fifteen years, this has included application security and system design for banking and government institutions, the architecture of a national bank's big-data platform and fraud-detection system, the founding and technical leadership of a start-up, and independent consulting in cloud, data and AI.
Keeping up with science is part of my routine. Currently pursuing my continuing education through a Master's in High Performance Computing Engineering at Politecnico di Milano, with research interests in scalable simulation systems, parallel computing, and emerging quantum machine learning paradigms.
Cloud, DevOps & MLOps
Private and public cloud infrastructure, CI/CD, automation and ML model delivery, with monitoring, high availability and cost efficiency built in.
Security
Application security, secure SDLC and system hardening, from penetration testing and code review to WAFs, honeypots and SIEM.
Performance & AI
Performance-oriented software in modern C++, CUDA and MPI, and applied AI from LLM agents to physics-informed and bio-inspired models.
Experience
Where I have worked
Fifteen years across security, big data, AI and my own companies, always hands-on.
Sep 2018 – Present
Milan / Tehran
Independent Research Engineer, Consultant & Start-up Founder
Self-employed · AI, cloud, data & HPC
- HPC and scientific computing (2025–present). I build parallel and GPU software in modern C++ with CUDA, OpenCL, OpenMP and MPI, and run it on CINECA's Galileo100 supercomputer with SLURM and Singularity/Apptainer. This includes wormsim2, a real-time C. elegans simulator about 40× faster than serial; AstraLog-HPC, a C++17 telemetry pipeline built as one full software-engineering cycle; and SC-PINN, a physics-informed graph neural network for ion-channel gating.
- Co-founder & CEO, Neur Asal Ware (2022–2023). I led architecture, delivery and hiring. I ran our cloud on OpenStack with GitLab CI/CD for e-commerce and real-estate clients and cut infrastructure cost by about 30%.
- Agentic AI. I built and run a multi-agent system on Claude: parallel sub-agents, custom MCP tools, structured outputs and automatic checks on every result. I use it every day for real work. Designed and built MatNeuron, an AI-powered web-based simulation environment for neural and biomechanical models.
- Freelance consulting. Cloud architecture, secure-SDLC reviews, data-platform design and scientific software.
- Start-up mentoring. I mentored teams building a portable point-of-sale platform for Day Bank, a platform for investing in property through fractional shares, a group-savings app, a web-shop builder in React, and a local blockchain platform on Stellar.
Apr 2015 – Sep 2018
Tehran
Big Data Platform Engineer & Data Scientist
Sadad Informatics Corp. · Melli Bank Iran
- Designed and rolled out an on-premise Hadoop/Spark platform on OpenStack from scratch. It replaced a costly proprietary stack and cut data-infrastructure cost about five times.
- Co-designed a near-real-time fraud-detection system that combines streaming and batch ML scoring on banking transactions.
- Coordinated two engineering teams, one for data infrastructure and one for data science.
- Consulted on secure authentication and device fingerprinting (a "digital DNA" of each user's sessions) and on a mobile-banking security SDK.
Jul 2008 – Mar 2015
Tehran
Software & Systems Engineer: application and system design
Tosan · Saba System Sadra
- System Engineer & Project Manager, Saba System Sadra (2012–2015). I designed Azaram, a Linux web-application firewall with HTTPS offloading, load balancing, high availability, a self-healing monitoring layer and a signature-update centre. It protected government and financial web systems. I managed the project and the team. I also worked on the Mavara SIEM and the Kandoo honeypot.
- Software Security Engineer & Trainer, Tosan (2008–2012). White-box and black-box penetration tests, secure-code reviews, fuzzing and reverse engineering for banking and government systems. I wrote security patches and secure-development policies.
- Training and consulting (2010–2014). I trained developers on the secure SDLC (OWASP, WASC and Microsoft SDL), penetration testing and code review, and ran security assessments for banks, telecom and energy companies and public bodies.
- Honeynet Project, Iranian chapter (2011). Volunteer researcher: honeypots such as Glastopf, and malware and rootkit detection.
My research roles (OpenWorm Foundation and University of Tehran) are in the Research section.
Selected work
Things I have built
A few systems I designed and built myself or together with my teams.
A bank's big-data platform
An on-premise Hadoop/Spark platform on OpenStack for a national bank. It replaced a proprietary stack, cost about 5× less, and two teams built on it every day.
Near-real-time fraud detection
Streaming and batch ML scoring on banking transactions, so suspicious activity is scored in near real time.
Azaram web-application firewall
A Linux WAF with HTTPS offloading, load balancing, high availability, self-healing monitoring and a signature-update centre, used by government and financial web systems.
Multi-agent AI workflow
A working multi-agent system on Claude with parallel sub-agents, custom MCP tools and structured outputs. Every result is checked automatically before I use it.
AstraLog-HPC
A C++17 telemetry anomaly-detection pipeline, built as one full software-engineering cycle: requirements, architecture, V&V, CI/CD, and Apptainer + SLURM runs on CINECA's Galileo100.
A start-up's own cloud
Our own OpenStack cloud with GitLab CI/CD for e-commerce and real-estate clients. It cost about 30% less to run than before.
Skills
Tools I work with
Software development lifecycle
End-to-end software engineering as one documented and traceable cycle: requirements, architecture, implementation, verification and validation, CI/CD, release and operation. Applied in full on AstraLog-HPC, from use cases to SLURM runs on a supercomputer, and earlier through secure-SDLC training for banking and government developers.
Languages & backend
AI & machine learning
Data
Cloud, DevOps & MLOps
HPC
Security
Research · 20% of my time
Science as part of the routine
Research has been a long-standing part of my week: about 20% of my time, including evenings and weekends, is dedicated to science. For more than ten years this has ranged from computational biology and neuroscience to mathematics, physics and robotics, with a long-term focus on bio-inspired AI and quantum computing.
My usual week
80% engineering work
20% research
From biology to math, physics and robotics, and a long-term vision of bio-inspired AI and quantum computing.
Biology & neuroscience
Ion channels and the nervous system of the small worm C. elegans. This is where my M.Sc. studies in biochemistry and my OpenWorm work began.
Mathematics
Models that fit real data: Hodgkin–Huxley and Markov models, numerical linear algebra and numerical methods for PDEs.
WormMathBookPhysics
Biophysics of cells and the mechanics of bodies: finite-element models, and neural networks that respect physical laws (PINNs).
Robotics & control
M.Sc. studies in Mechatronics: control, signals and systems, neural networks. Today I simulate a worm's nervous system and body together.
Bio-inspired AI
How a small nervous system makes behaviour, and what this can teach the next AI models. MatNeuron is a web-based simulation environment that bridges neural networks with biomechanical systems.
MatNeuronQuantum computing
My long-term vision: the next generation of scalable systems will join HPC and quantum computing, and open the way to quantum machine learning and new neuro-inspired AI. I study HPC as the first step.
Selected research work
-
OpenWorm Foundation · Senior Contributor (volunteer)
Nov 2014 – PresentI proposed and led ChannelWorm (2015–2018): I curated patch-clamp data from papers, fitted Hodgkin–Huxley models, estimated kinetics for channels with no data, and built the Python APIs and validation tests used across OpenWorm.
ChannelWorm -
wormsim2
2025 – PresentA real-time C. elegans simulator: a Hodgkin–Huxley neural model joined to a finite-element body. The C++ core has CUDA, OpenCL and OpenMP back-ends, runs about 40× faster than serial, and runs on CINECA's Galileo100 with SLURM.
3D viewer Code -
chModeler & Chopen
2018 – PresentA compact ion-channel model (about 5 parameters) that fits a curated set of about 270 channels in seconds, with an ML predictor of channel kinetics from the amino-acid sequence. A proof of concept, still in development.
chModeler Chopen -
SC-PINN
2026 · Politecnico di MilanoA physics-informed graph neural network in PyTorch, with ESM2 protein-language-model embeddings, that predicts how ion channels open and close.
-
Research Assistant · University of Tehran
Mar – Sep 2014Lab of Systems Biology and Bioinformatics: I predicted ion-channel kinetics from protein sequence with ML and sequence alignment, and built a dataset of more than 200 channels from the literature.
-
Smart pacemakers & ECG
2019 – 2020I set up and mentored a student team from different fields to study smarter pacemaker and ECG algorithms, using Python and PhysioNet data.
Education
Education & certificates
Degrees
M.Sc. High Performance Computing Engineering
2025 – PresentPolitecnico di Milano, Milan
Parallel Computing (CUDA/OpenMP), Advanced Methods for Scientific Computing (modern C++), Numerical Methods for PDEs, Numerical Linear Algebra, Numerical Analysis for ML, Applied Statistics, Computing Infrastructures, Software Engineering for HPC.
M.Sc. studies, Mechatronics Engineering
2019 – 2021Sharif University of Technology, Kish campus
Coursework completed: artificial neural networks, advanced automatic control, signals and systems, dynamics.
M.Sc. studies, Biology / Biochemistry
2011 – 2014Payame Noor University, Tehran
Coursework completed. Thesis topic (not defended): Computational Modeling of Ion Channels and Structure-Function Relationship Analysis.
B.Sc. Information Technology Engineering
2005 – 2010Islamic Azad University, South Tehran
GPA 17.43/20, among the top graduates of my year.
Certificates
- Google Cloud Skills Boost (2022–2026): Vertex AI, MLOps, BigQuery ML, Generative AI with Gemini, Responsible AI. Badges on Credly
- IBM Data Science Professional Certificate (2021)
- Linux Foundation (2018): Introduction to Cloud Infrastructure Technologies; Blockchain for Business with Hyperledger
- Coursera (2015–2019): Python for Genomic Data Science (Johns Hopkins), Computational Neuroscience (University of Washington), Practical Machine Learning, Pattern Discovery in Data Mining, Front-End Web Development with React
Languages
- EnglishC1 · IELTS 7.0
- PersianNative
- ItalianLearning
- FrenchLearning
Contact
Let's talk
I am open to full-time roles in Milan (on-site or hybrid) and remote roles in Europe, in software, data, AI/ML and HPC engineering. I am also happy to talk about research and open-science projects.