Open to opportunities
MLOps Engineer Student
I build, deploy, and maintain production-ready Machine Learning and Generative AI systems, with a background in cybersecurity.
01
I'm an MLOps Engineering student who enjoys building things and figuring out how they work.
My interests are mainly around machine learning, AI and infrastructure, but I like working across different areas rather than sticking to one part of the stack. I've worked on everything from forecasting and LLMs to cloud services and small hardware projects.
Before moving into MLOps, I worked in cybersecurity for a few years. That background still influences how I approach building and understanding systems.
02
Featured · Sold to a client
Licensed desktop application (PySide6) that I designed, built, and sold to a company in Turkey. It automates airline ticket processing: PDF/OCR parsing, vendor-agnostic LLM field extraction, exchange-rate-based billing, and automated airline booking verification with Playwright. Ships with offline signed licensing and signed auto-updates, with update delivery and patching infrastructure on AWS.
Offline edge AI voice assistant. A Raspberry Pi Pico 2 W streams audio over Wi-Fi to a FastAPI backend on a Raspberry Pi 5 running a fully local pipeline: whisper.cpp speech-to-text, quantized Qwen 2B via llama.cpp, and Piper text-to-speech, with MLflow monitoring.
Clinical decision-support research prototype using LangChain and OpenAI, with human-in-the-loop doctor review and optional ICD-11 enrichment.
Low-cost, always-on home surveillance system on a Raspberry Pi with ML-based face recognition, real-time alerting with snapshot capture, deployed as a 24/7 Linux service with GitHub Actions CI/CD.
AI onboarding copilot using retrieval-augmented generation (RAG) with escalation levels and MLflow-based LLMOps for experiment tracking and evaluation.
IoT room-pet that turns environmental and system data into moods, expressions, and feedback. A Raspberry Pi Pico streams sensor data over MQTT to a Python backend with TimescaleDB and Grafana; optional LLM personality responses are traced with MLflow. Deployed on Azure with Docker Compose and CI/CD.
Production-like FastAPI application on AWS EC2 with Docker and PostgreSQL, automated server setup via Bash, monitoring, backups, and NGINX-based API exposure.
03
Stockholm
Stockholm
Stockholm
Fairfax, VA
Nackademin
MLOps Engineering
2025 – 2027
Stockholm University
BS, Philosophy, Economics & Politics (courses completed)
2020 – 2023