Istanbul, Türkiye
ML Inference Systems Engineer at Trendyol, where I build the platform serving 1M+ requests per minute to 100M+ customers — recommendation, search, and increasingly AI agents in production. My work sits at the intersection of ML, systems, and infrastructure: vLLM and Triton for serving, Kubernetes across cloud and on-prem, and Piper, our in-house orchestration system for the deployment lifecycle. I care about the parts of ML that don't make it into most conference talks — p99 latency, GPU utilization, what breaks at 3am, and how inference systems actually live once they're deployed. I co-organize GDG Istanbul, where we run monthly meetups and DevFest — one of the larger developer conferences in the region. I speak regularly at Build with AI, DevFest, and GDG events across Turkey and abroad (most recently GDG Baku). I contribute to Google's Agent Development Kit and write about production AI agents, inference optimization, and ML platform engineering. If you work on any of this, say hi.
- Building and developing internal RAG LLM tools with internal documentation enhancing the productivity and efficiency of Trendyol developers to reach to a specific information in the internal documentation, QA platform, and communication platform. - Optimizing and Deploying large scale ML models with KServe and Nvidia Triton which directly impact the user application on various domains such as search and NLP. - Maintaining stable Experimentation Environments of Data Science team with Kubeflow on GCP for better development experience and increased productivity.
- Assisted in building the backend and DevOPS of an automated machine learning platform for corporate customers using django rest framework, AWS Autogluon, Openshift, jenkins, kubernetes and S3 object storage, where users can upload their datasets and train machine learning models with automated process that includes data preprocessing, automated feature engineering, automated splitting, training phase, testing phase and result phase where the best model is presented. - Participated in building a data insights platform where users could have important insights about their customers from segmentation to data visualization with various properties using SQL queries and FastAPI to create endpoints of the query results for the users. - Assisted in creating machine learning models that trained on the cooling systems in data centers to present efficient and optimized solutions that result positively on the budget for cooling data centers. - Created personalized data insights that fit the customers’ needs and requirements from big data platforms using Oracle SQ