Skip to content
View yigitliman's full-sized avatar
  • Technical University of Munich
  • Munich
  • 23:27 (UTC +02:00)
  • LinkedIn in/yigitliman

Block or report yigitliman

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
yigitliman/README.md

Yigit Liman

AI Master's Student @ FAU · IT Specialist @ TUM

LinkedIn Location Open to work

AI Platform & HPC Infrastructure Engineer

M.Sc. thesis at NHR@FAU on eBPF-based NFS rate limiting for HPC clusters


Tech

Core: Python · Linux · eBPF Platform: Kubernetes · Terraform Serving: FastAPI · Docker · Docker Compose Orchestration: Apache Airflow CI/CD: GitHub Actions · GHCR Monitoring: Prometheus · Grafana ML: PyTorch · XGBoost · scikit-learn · MLflow · Evidently LLM / RAG: LangChain · ChromaDB · Anthropic Claude


Projects

Infrastructure

Project What it does Stack
ai-platform-lab End-to-end AI platform: Terraform-provisioned Kubernetes with remote state, monitoring and a CI/CD pipeline shipping containers to the cluster Terraform, Kubernetes, GitHub Actions, Docker
gpu-scheduler Fair-share GPU cluster scheduler with EASY backfill and starvation-free job aging Python, FastAPI, Prometheus, Grafana, Docker
llm-inference-server Continuous-batching inference server with iteration-level scheduling and TTFT/throughput metrics PyTorch, Transformers, FastAPI, Prometheus, Docker

Robotics

Project What it does Stack
pepper-games Voice games, tablet kiosk and web admin panel for a SoftBank Pepper robot, answered by a self-hosted LLM Python, naoqi SDK, Whisper, gpt-oss-120b

MLOps

One shared stack across the series, where each project adds a piece the previous one did not have.

Project What it does Stack
churn-prediction The baseline of the series: a model registry and drift detection wired into a serving API XGBoost, MLflow, Evidently, Prometheus, Docker, K8s
price-prediction Same pipeline pointed at a regression target, so the metrics are errors rather than class scores XGBoost, MLflow, Evidently, FastAPI, Docker
anomaly-detection Drops the labels entirely and scores sensor readings with an unsupervised model IsolationForest, MLflow, Evidently, FastAPI, Docker
airflow-ml-pipeline Takes training off my machine into a scheduled DAG that retrains when drift crosses a threshold Airflow, MLflow, Docker Compose
image-classifier First GPU-trained deep learning model here, a CIFAR-10 CNN behind the same API shape PyTorch, MLflow, FastAPI, Docker
document-qa Answers from retrieved documents instead of a trained model, and evaluates the retrieval itself LangChain, ChromaDB, Claude, FastAPI, Docker
mlops-sentiment-pipeline Splits the single container into a multi-service Compose stack with a transformer doing inference DistilBERT, MLflow, Evidently, FastAPI, Docker Compose

Every MLOps and Infrastructure project ships with tests, a Dockerfile, and a GitHub Actions pipeline.

Popular repositories Loading

  1. yigitliman.github.io yigitliman.github.io Public

    Personal CV page

    HTML

  2. image-classifier image-classifier Public

    CIFAR-10 image classifier API: PyTorch, MLflow, FastAPI, Docker, CI/CD.

    Python

  3. churn-prediction churn-prediction Public

    Customer churn prediction API: XGBoost, MLflow tracking and registry, Evidently drift monitoring, Prometheus/Grafana, FastAPI, Docker, CI/CD.

    Python

  4. price-prediction price-prediction Public

    California house price prediction API: XGBoost, MLflow, Evidently, FastAPI, Docker, CI/CD.

    Python

  5. anomaly-detection anomaly-detection Public

    Sensor anomaly detection API: IsolationForest, MLflow, Evidently, FastAPI, Docker, CI/CD.

    Python

  6. airflow-ml-pipeline airflow-ml-pipeline Public

    Airflow MLOps pipeline: scheduled training, drift-triggered retraining, MLflow tracking, Docker Compose.

    Python