Ogulcan Aydogan DevOps Engineer in London

Ogulcan Aydogan

DevOps, Cloud & AI Engineer
Phone

+44 7355 005050

Location

London,
United Kingdom

Projects

I build open-source AI/ML tools, train large language models for low-resource languages, and develop production AI systems. Below are my AI & Machine Learning projects alongside production web applications running on my own AWS infrastructure.

AI & Machine Learning

MCP Security Scanner

Security scanner for Model Context Protocol (MCP) servers — detects prompt injection, tool poisoning, privilege escalation, and cross-tool attacks.

Python AI Safety MCP Security

EU AI Act Compliance Kit

Automated compliance assessment for EU AI Act (2024/1689) — classifies AI systems by risk tier and generates audit-ready compliance reports.

Python Compliance EU AI Act Pydantic

Turkish LLM Benchmark

Standardised evaluation suite for Turkish language models — 500+ expert-curated questions across reasoning, math, cultural knowledge, translation, and coding.

Python NLP LLM Eval Turkish

LLM Behavior Diff

Semantic diff tool for LLM upgrades — detects behavioral regressions, safety boundary changes, and hallucination shifts between model versions.

Python PyTorch MLOps Testing

Agent Memory Sentinel

Real-time security monitoring for AI agent memory — detects poisoning, injection attacks, and anomalous behavioral drift with Prometheus metrics.

Python AI Safety ChromaDB Async

vLLM Production Stack

Kubernetes-ready production deployment framework for vLLM with multi-GPU scheduling, autoscaling, monitoring, and security policies on AWS EKS.

Kubernetes Helm Terraform vLLM

DGX Spark LLM Stack

Pre-built PyTorch/Triton/flash-attention wheels optimised for NVIDIA DGX Spark (Blackwell GB10) with reproducible build scripts and benchmarks.

PyTorch CUDA Triton NVIDIA

Yapay Voice Assistant

Full Turkish-language enterprise voice AI platform with Whisper v3 STT, XTTS v2 TTS, and integrated Turkish LLM for end-to-end conversation.

Whisper XTTS FastAPI LLM

Rendera

AI-powered 3D room rendering platform with geometry-locked Blender/Cycles pipeline, FLUX style transfer, and Stripe subscription billing.

FastAPI Next.js Blender FLUX

Sports Probability Engine

Multi-sport ML analytics platform with ensemble predictions (Poisson + Gradient Boosting + Neural Net), live odds comparison, and value bet detection.

PyTorch FastAPI Next.js ML

Algo Trading Lab

Modular trading bot framework with EMA/RSI signal generation, macro sentiment engine, multi-asset portfolio support, and AI prediction endpoints.

Python CCXT PyTorch FastAPI

Turkish LLM Family

Fine-tuned Turkish language models (1.5B–72B) published on HuggingFace — trained with SFT and DPO on multi-GPU clusters for Turkish NLP tasks.

HuggingFace SFT DPO A100/V100

Side Projects & Web Apps

Production web applications on my own AWS infrastructure — each with CI/CD via Jenkins, custom subdomain, S3 hosting, and CloudFront CDN.

RSVP Manager

rsvp.ogulcanaydogan.com

Event RSVP platform for creating, sharing, and tracking invitations with real-time guest responses.

Python Flask S3 CDN

Bash Prompt Customizer

bash.ogulcanaydogan.com

Design custom terminal prompts with vivid colors, live preview, and one-click export for Bash, Zsh, and PowerShell.

JS CSS S3 CDN

PricePulse

price.ogulcanaydogan.com

Family price tracking hub — monitor products across retailers, set price drop alerts, and share watchlists with your household.

Python Lambda Dynamo CDN

Döviz Radarı

doviz.ogulcanaydogan.com

Live Turkish lira exchange rates, gold and silver prices, 7-day trend charts, and a quick currency converter.

JS Chart.js REST API S3

GoPro Live Stream

live.ogulcanaydogan.com

Private HLS video streaming player for GoPro cameras with real-time chat, stream health metrics, and Picture-in-Picture support.

HLS WebSocket JS S3

Unit Conversion Hub

convert.ogulcanaydogan.com

Convert units, compare time zones, and calculate elapsed days — a lightweight toolkit built for cloud practitioners.

JS HTML5 S3 CDN

Infrastructure Overview

All six projects share a unified AWS infrastructure stack that I built and maintain:

Hosting & CDN

AWS S3 static hosting with CloudFront CDN, custom SSL certificates via ACM, and Route 53 DNS management for all subdomains.

CI/CD Pipelines

Each project has a dedicated Jenkins pipeline with automated linting, testing, S3 deployment, and CloudFront cache invalidation on every push.

Security

HTTPS enforced across all subdomains, IAM least-privilege policies, S3 bucket policies with OAI, and automated security headers via CloudFront functions.

Monitoring

CloudWatch metrics and alarms for all services, S3 access logging, CloudFront real-time logs, and cost anomaly alerts.