Software Engineer building product systems end-to-end — web, AI, and infrastructure — and shipping them. Full-stack in TypeScript + Python, with a growing focus on ML/MLOps and Kubernetes reliability. Started in commerce, self-taught the modern stack, now shipping across a multi-product portfolio.
| Features end-to-end — React / Next.js / Svelte on the front, Node / FastAPI on the back, Postgres / Mongo underneath. Design → API → UI → deploy. | Failure prediction, calibrated models, in-cluster serving, drift-triggered retraining — plus AI-driven K8s incident response and auto-remediation. | BLE protocols and ESP32 firmware in C, Terraform / Helm / ArgoCD infrastructure, observability with Prometheus & Grafana. |
| Project | What it is | Stack |
|---|---|---|
cluster-canary
|
Predicts Kubernetes pod failures (OOMKill, CrashLoopBackOff) ~30 min ahead — calibrated probabilities + SHAP, served as an in-cluster gRPC sidecar (p95 < 50 ms). | Python · LightGBM · PyTorch · BentoML · Prefect · K8s |
openwrist
|
Open-source ESP32 smartwatch that speaks iPhone natively over BLE (ANCS/CTS/AMS) — notifications, calls, time, music, with no App Store app. | C · ESP-IDF · LVGL · SwiftUI · CoreBluetooth |
kubeai-ops
|
AI incident response for Kubernetes — detects issues, runs multi-LLM root-cause analysis, auto-remediates with ChatOps + RBAC. | Python · Svelte · Terraform · Multi-LLM |
| RestaurantOS | Full-stack restaurant management — orders, menu, kitchen, billing. Dockerised for one-command deploys. | JavaScript · Node.js · Docker |
Languages
Frontend
Backend
AI / Data
Data
Infra & Cloud
Shipping full-stack + AI features, and building cluster-canary and openwrist in the open. Focused on K8s reliability, MLOps, and the details that separate a demo from a system.
design → api → ui → deploy → observe