diff --git a/README.md b/README.md
index 1215bd5..c9e37d1 100644
--- a/README.md
+++ b/README.md
@@ -1,22 +1,88 @@
-# About Me:
-I build infrastructure that runs itself and automation that replaces repetitive work. My repos are a mix of cloud-native tooling re-implementations in Go, AI-powered automation
pipelines in Python, and full-stack SaaS products in TypeScript. I run a home lab with
35+ Docker services across multi-region clusters because production is the best teacher.
+Usman Mustafa
+Staff DevOps / SRE / Platform Engineer
+
+12+ years building infrastructure that runs itself — Kubernetes platforms, HPC clusters, and the automation that keeps both quiet at 3 AM.
+
-## Socials:
-[](https://linkedin.com/in/usmanmustafa4u2) [](https://x.com/Dxbusman)
+
+AWS Certified Solutions Architect – Professional ·
+Kubernetes & Helm ·
+ArgoCD / GitOps ·
+HPC (LSF / SLURM) ·
+Observability & SLOs
+
-# Tech Stack:
-                                                                                          
-# GitHub Stats:
-
-
-
+---
+
+## What I Do
+
+**Platform & SRE**
+I design and run Kubernetes platforms end to end — cluster lifecycle, Helm charts, GitOps delivery with ArgoCD, and the guardrails (policy, autoscaling, cost control) that let product teams ship without paging the platform team. I think in SLOs and error budgets, not just uptime.
+
+**HPC & GPU Infrastructure**
+Before "platform engineering" had a name, I was tuning job schedulers. I've operated IBM LSF clusters at 1,700+ node scale and work regularly with SLURM and GPU/CUDA workloads — the queueing, fair-share, and scheduling problems most cloud-native tooling never has to solve. That background carries directly into Kubernetes-based ML/GPU platforms, where the same instincts apply to GPU bin-packing and inference serving.
+
+**AI-Augmented Operations**
+I build the automation layer on top of the infrastructure layer: Claude Code and agent-based workflows for operational toil, n8n for orchestration, and infrastructure-as-code so automation stays reviewable and reproducible instead of a pile of one-off scripts. Observability (ELK, Zabbix, Prometheus) is what makes any of this safe to automate — you can't automate what you can't measure.
+
+---
-### Random Dev Quote
-
+## Featured Projects
-### Top Contributed Repo
-
+Two independent, from-scratch portfolio projects applying the same idea — an adaptive, metrics-driven control loop — to two different infrastructure domains:
+
+### [kube-ai](https://github.com/vjranagit/kube-ai)
+Adaptive vLLM/GPU serving tuner for Kubernetes. A SelfTune-style control loop that auto-tunes replica count and `vLLM --max-num-seqs` from live metrics, with Prometheus/Grafana observability and a web UI.
+`Python` · `Kubernetes` · `vLLM` · `Prometheus` · `Grafana` · `MLOps`
+
+### [slurm-ai](https://github.com/vjranagit/slurm-ai) — `adaptive-wlm-controller`
+Quota-aware adaptive control loop for Slurm clusters. Reads scheduler queue pressure, computes a saturation score, and safely tunes job-concurrency limits (AIMD by default, optional tabular Q-learning) via `sacctmgr`/`scontrol` — dry-run by default, Prometheus-instrumented, 180+ tests passing. Inspired by the SelfTune paper (USENIX NSDI '23). IBM LSF support is a planned follow-on.
+`Python` · `Slurm` · `HPC` · `Prometheus` · `Reinforcement Learning`
+
+### Also on GitHub
+- **[zabbix-mcp-server](https://github.com/vjranagit/zabbix-mcp-server)** — Lightweight MCP server wrapping `zabbix-cli` with 24 AI-accessible tools, so Claude Code and other AI assistants can query and act on Zabbix monitoring data.
+- **[mcp-servers-unified](https://github.com/vjranagit/mcp-servers-unified)** — A unified collection of MCP server implementations (Gmail, Zabbix, ELK) plus the Kubernetes/Ansible/Docker deployment infrastructure behind them.
+
+*(Plus 15+ from-scratch Go re-implementations of CNCF ecosystem tools — Argo Workflows, Crossplane, Volcano, Tekton, and others — built to understand the internals, not to replace the originals.)*
---
-[](https://visitcount.itsvg.in)
+
+## Tech Stack
+
+**Cloud & Platform**
+     
+
+**HPC & GPU**
+   
+
+**Observability**
+   
+
+**Automation & IaC**
+   
+
+**Languages**
+ 
+
+---
+
+## GitHub Stats
+
+
+
+---
+
+## Let's Connect
+
+
+
+- Open an issue on any of my repos — I check them.
+- LinkedIn / X / Email / Website: *add your links here before publishing*