Charis Nikolaidis | Software Engineer

Career 💡

I'm Charis, a Software Engineer in Athens, Greece. I build enterprise systems and multi-tenant platforms, mostly Python and C#. I keep the infrastructure intentionally boring because I'm fond of sleeping through the night. A system that needs a 2 AM pager call to stay alive isn't architecture, it's a hostage situation.

Full Stack 🖥️

Backend is where things make sense, but I ship frontend too (React or Angular, depending on who reached the repo first). I have a soft spot for the unglamorous wins: refactoring a query that reads like a crime scene, or getting an API to answer before the user gives up and hits refresh. The goal is removing surprises, not adding buttons.

Innovation 🔧

Most of what I build leans on automation or AI, not to chase a trend, but because I'm fiercely protective of my own time. I'm aggressively slow to add complexity: an abstraction has to prove it repeats itself before it earns a place in the stack. Honestly, half the job is talking people out of over-engineering a CRUD app.

Security 🔐

Security is invisible right up until it's catastrophic. I have an MSc in Computing (Internet Technology & Security), which mostly handed me a lifelong case of professional paranoia. I focus on the unsexy layers that actually hold: strict RBAC, audit logs that survive the post-mortem, and storing secrets literally anywhere except the codebase. I overthink this part on purpose, so nobody else has to.

Automation 🤖

If I have to do a task by hand three times, the fourth time it's a script's problem. Call it discipline; it's really just a deep allergy to chores. Cron jobs and background workers never complain, never get distracted, and have never once made a typo at 5 PM.

Integrations 🔗

A surprising amount of real engineering is just convincing two systems that clearly despise each other to share data. Payments, third-party APIs, a legacy database with deep trust issues. The hard part is never the happy path; it's the undocumented error response that only ever surfaces on a Friday afternoon.

Privacy 🕵️‍♂️

I treat user data like radioactive material: collect as little as possible, encrypt the rest, and keep a very short list of who's allowed near it. Compliance is mind-numbingly dry, right up until the one afternoon it quietly prevents a corporate existential crisis.

Meet My Work 🏗️

Here's what I've actually shipped to production, plus a few weekend builds I made purely for the technical dopamine. If it was a 48-hour hack, I'll say so. If it was a year of production work and the awkward trade-offs that come with it, that's in here too. Have a look.

Contact

If you're building a system, automations, security heavy systems, planning for a migration, or even up for debate on why AI will take my job, reach out.

Europe, Greece
UTC+2

What I've built

Enterprise systems first, then smaller work kept honestly sized.

Maritime supply chain platform

2025 to now Enterprise system

A multi-tenant maritime supply chain platform, built from an empty repo and owned end to end. Async FastAPI on PostgreSQL and Redis, live AIS vessel tracking, procurement, CRM and accounting, with the tracing and audit trails needed to keep the thing honest. It is the largest system I've built and the one I'll defend line by line.

Built with:

Python
FastAPI (async)
SQLAlchemy
PostgreSQL
PostGIS
Redis
Celery + Workers
OpenTelemetry
Prometheus
Docker
Azure

What's inside:

Layered async FastAPI on PostgreSQL and Redis, with clear service boundaries
Multi-tenant, with strict per-account isolation, RBAC and field-level audit logging
TDD-driven development with around 3570 tests, DRY and SOLID principles
Design patterns used: Repository, Unit of Work. Creational patterns like Factory, Builder. Behavioral patterns like Observer, Strategy. Structural patterns like Adapter, Decorator.
Geospatial vessel and route data on PostGIS, more than a billion points across thousands of vessels
A full procurement flow: jobs, supplier inquiries, quotations and purchase orders, with CRM and accounting around it
Per-request transaction boundaries with explicit commit and rollback handling
AI-assisted workflows with context-aware draft replies for end users
Full-text search with GIN indexes, async background jobs on Celery, traced with OpenTelemetry and Prometheus
Built from an empty repo and still actively developed in production

Legacy system rescue

2025 to now Enterprise system

Inherited a database first legacy system that two engineers had spent a year failing to stabilise. It was a mess. I refactored the backend, rebuilt the front end, and had it in production in about three months. ASP.NET Core and Angular, a few hundred endpoints, a pile of legacy decisions quietly corrected. Nobody wrote a press release. It just started working.

Built with:

C#
ASP.NET Core 8
Entity Framework Core
SQL Server
Angular 19
TypeScript
Azure

What it took:

Took over a stalled, database-first codebase and shipped it to production in about three months
Full front-end rebuild in Angular, backend refactor across roughly ninety controllers and a hundred services
Multi-role access control with per-user scoping, and Save, Approve, Submit workflows across every document
PDF and Excel generation, Azure Blob storage and Application Insights telemetry
Stabilised the data layer, the stored procedures and the deployment pipeline
Still running in production, after a few hundred commits since I took it over

Gpot

2026 to now SaaS, live

gpot.gr. A credit control SaaS for Greek businesses. The hard part is the sync: it pulls invoices and payments from myDATA, the Greek e-invoicing platform run by the tax authority, and from the customer's ERP, then reconciles tax, ERP and manual records before anyone goes chasing a debt. FastAPI and React, Stripe for its own billing, multi-tenant.

gpot.gr

Built with:

Python
FastAPI
React
TypeScript
PostgreSQL
Redis
Celery + Workers
Stripe
myDATA (AADE)

The hard part is the sync:

Pulls from myDATA, Greece's e-invoicing platform, and from any ERP, through a small connector that reads the customer's database and exposes one standard API
Reconciles invoices and payments across three sources, myDATA, ERP and manual, and flags conflicts instead of overwriting them
Celery workers on dedicated queues, syncing many tenants at once on per-account schedules
Resumable syncs with cursor pagination, time budgets and idempotent upserts, so a killed run picks up where it stopped
Per-account rate limiting and locks so two syncs never race on the same data
Encrypted credentials, Stripe billing, Advanced Multi-source Sync and 2FA

Greek web scraper

2024 Published package

A Scrapy based crawler for Greek-language web content, with NLP filtering to drop the junk and clean JSONL on the way out. Built for a data collection job, then packaged and pushed to PyPI in case anyone else needs a pile of Greek text. So you can just pip install it.

GitHub · PyPI

Built with:

Python
Scrapy
spaCy
NLP filtering
JSONL output

What it does:

Concurrent crawling on Scrapy
NLP filtering to keep Greek content and drop boilerplate
Clean JSONL output for downstream training
Retry and encoding handling for messy pages
Published on PyPI with docs

sarcastic_ai_agent

2025 Weekend experiment

A weekend experiment. I fine-tuned Gemma 3 1B with QLoRA to be reliably sarcastic, then pointed it at my own code for feedback I didn't ask for. Not a product, never going to be one. It's here because it covers the whole fine-tuning loop, dataset to adapter to inference, on a single GPU.

Built with:

Python
PyTorch
Transformers
PEFT (LoRA / QLoRA)
TRL
Gemma 3 1B

What it is:

QLoRA fine-tune with 4-bit and 8-bit quantization on one GPU
A small hand-built sarcastic chat dataset in JSONL
LoRA adapters across the projection layers
An interactive script to chat with the result
Built for fun, kept small, and useful mainly as a complete fine-tuning exercise.

Image recognition

2024 Learning project

A small Flask app that recognises objects in uploaded images using OpenCV and a pretrained model. Nothing groundbreaking. I built it to get the end to end shape of a vision pipeline, upload, infer, return a label, rather than to chase the state of the art.

GitHub

Built with:

Python
Flask
OpenCV
Pretrained CNN

What it does:

Upload an image, get classified objects back
Uses a pretrained model through transfer learning
A simple Flask interface
Built to learn the pipeline, kept small on purpose

FuelStation ERP

2022 Where it started

My first real system. An ERP for running a fuel station, written in C# and .NET with a Blazor and WinForms front. It's not how I'd build it today, and that's the point: it's where I learned what a data model wants, what happens when you skip validation, and why the boring parts matter. Still on my GitHub, warts and all.

GitHub

Built with:

C#
.NET
Blazor
WinForms
SQL Server

What it was:

Stock, sales and pricing for a fuel station
Both a desktop and a web front end
My first proper data model and the lessons that came with it
Two GitHub stars, which I'm unreasonably fond of

MSc security capstone

2024 Academic

The build for my MSc in Computing (Internet Technology & Security). A web application in PHP where the security work was the actual subject: input handling, authentication, the common web vulnerabilities and how not to ship them. Academic, but it's where the security habits I use now got their formal grounding.

GitHub

Built with:

PHP
MySQL
Web security

What it covered:

Threat modelling and the OWASP top ten in practice
Authentication and access control done deliberately
Written up as the MSc dissertation project
Formal grounding for the security habits I still use
University of Northampton, 2023 to 2025

Homelab network segmentation

2025 Homelab

My home network, taken more seriously than a home network needs to be. Separate VLANs for servers, admin, IoT and guests on a UniFi Dream Machine Pro, with firewall rules between them so the smart plug can't reach the things that matter. This is also where the two years of networking, the VLANs and VPNs and the Cisco Packet Tracer coursework behind it, actually live.

Built with:

UniFi Dream Machine Pro
VLANs (802.1Q)
Firewall rules
IDS/IPS
VPN

What's set up:

VLAN isolation for servers, admin, IoT and guest
Firewall policies controlling traffic between segments
IDS/IPS and traffic logging
Grew out of Cisco Packet Tracer coursework and two years of network experience

Charis Nikolaidis, Software Engineer in Athens, Greece

Career 💡

Full Stack 🖥️

Innovation 🔧

Security 🔐

Automation 🤖

Integrations 🔗

Privacy 🕵️‍♂️

Meet My Work 🏗️

Contact

What I've built

Built with:

What's inside:

Built with:

What it took:

Built with:

The hard part is the sync:

Built with:

What it does:

Built with:

What it is:

Built with:

What it does:

Built with:

What it was:

Built with:

What it covered:

Built with:

What's set up:

Technical Glossary