Project
Built a production-style LLM inference platform with KServe, vLLM, llm-d, Envoy Gateway, Argo CD, and Helm, exploring scalable model serving, request routing, and GitOps workflows.
Project
A Kubernetes operator that analyzes Prometheus metrics during rollouts to automatically detect regressions and enable safer, metrics-driven deployments.
Project
An intelligent autoscaling system that forecasts workload demand and proactively scales Kubernetes applications using KEDA and time-series predictions.
Project
Automated provisioning of isolated MedusaJS stores using Argo Workflows, enabling repeatable deployments with GitOps and cloud-native infrastructure.
Project
A seven-day build log exploring GPU discovery, KV cache management, batching, and request routing for AI workloads.
Project
A practical walk-through of the MLOps lifecycle, from version control and training data to serving, monitoring, and iteration.