• Developed a Kubernetes test environment on GKE + Terraform to run web app across pods with autoscaling, health/ready checks, secure ingress; integrated Prometheus/Grafana; leveraged underutilized on-prem capacity to reduce cloud spend.

• Shipped a production customer support chatbot (local Llama LLM on NVIDIA GPU + ChromaDB RAG with embeddings + web UI) enabling self-serve license lookup/reset workflows; deployed via Triton Inference Server with TensorRT-LLM backend and advanced inference tuning (quantization, KV-cache/attention optimization, continuous batching) to improve latency/throughput, deflecting 15+ tickets weekly and saving ~8–12 hours/week of manual resets.

• Cut cloud infrastructure cost by ~28% YoY by right-sizing EC2, tightening caching, enabling S3 lifecycle policies, and decommissioning idle resources.

• Led Tier-3 troubleshooting across Windows/Linux, Tomcat, Oracle/SQL Server, AD/SSO, drove root-cause fixes that cut repeat incidents ~25%.

• Improved report/query runtimes ~30% through index design, query-plan analysis, and targeted caching; eliminating most critical peak-time timeouts.

• Replaced recurring client workflows with Python automations (APIs/schedulers), boosting turnaround and avoiding tens of thousands in labor annually.

• Spearheaded the modernization of record-keeping systems, improving accountability and transparency for management.

• Centralized IT for ~100 users across multiple sites (Windows Server/AD, Azure AD/O365, Cisco/Meraki, VPN, MDM), reducing downtime ~20% and faster onboarding/offboarding.

• Digitized dispatch & scheduling with live dashboards (PowerShell/SQL + Monday, Geotab, Omnitracs, Samsara); reducing manual coordination ~30%, boosting on-time performance and safety features.

• Built 24/7 incident response with runbooks and clear escalation paths, reducing MTTR by 35%.

• Led safety initiatives and conducted training sessions to promote compliance and workplace safety, reducing recorded incident rates by ~20% in three months.

• Ran Linux on AWS EC2 with cost-aware VPC/site-to-site VPN, routing, and security groups, reducing hosting spend ~25% while meeting SLAs.

• Hardened and tuned sites (nginx, HTTPS/HSTS, caching, JS/CSS), improving speed and conversions.

• Led marketing experiments using an A/B testing framework with analytics, reducing CPC ~18% and conversion rate up on key campaigns.

• Built Python automations for customer comms (email/SMS/social APIs), reducing response times ~50% and eliminating manual queue backlogs; content/SEO ops drove ~20% YoY growth in online inventory revenue.