Case Studies
Projects & Case Studies
Impact-driven projects across cloud, AI, automation, and security — each told as Challenge → Solution → Impact.
Cloud & DevSecOps
B2B Commerce Platform: Cloud Build-out & Production Launch
A B2B retail distributor needed its CRM, mobile-app APIs, and legacy data moved onto a secure, observable cloud platform.
Built dev/staging/prod on AWS (RDS PostgreSQL, S3, ECR, CI pipelines), ran a phased legacy data migration, added daily AWS Backup, cAdvisor/Loki/Grafana monitoring, and Metabase dashboards exposed to AI tools via MCP behind Cloudflare Access.
Production launch with post-launch support; a security audit surfaced 24 findings (4 critical) with a prioritized fix plan; idle cloud resources cleaned up to cut spend.
AWSRDSECRCI/CDGrafanaLokiMetabaseSecurity Audit
AI & Automation
Self-hosted AI Assistant + WhatsApp Field-Ops Bot
A construction group’s site teams needed quick answers and simple daily reporting—without sending company data to public AI services.
On-prem LLM assistant (vLLM on GPU, Qwen2.5-14B) with RAG over company handbooks inside the team chat, plus a Twilio WhatsApp bot that syncs staff KYC onboarding, payment and material requests, and daily site updates into ERPNext.
Standardized field reporting across sites in Hindi, Hinglish, and English, with sensitive data kept in-house.
vLLMRAGLLMNode.jsTwilioERPNextMattermost
ERP & HR Automation
Group-wide HRMS on ERPNext / Frappe HR
Five group companies across several office locations needed one HR system with reliable attendance and payroll.
Frappe HR with biometric attendance sync, geo-fenced check-in, leave policies and approvals, payroll with salary-slip emails, per-company branding, and an employee self-service portal.
One HR platform for 5 companies; monthly payroll run end to end from a single system.
ERPNextFrappe HRBiometric SyncPayrollAutomation
Platform & Security
Self-hosted Internal Platform & Security Tooling
Cut hosting costs while getting visibility and control across the company’s internal services.
Moved team chat from EKS to an on-prem server; set up Prometheus/Grafana/Loki/Blackbox monitoring with chat alerting; self-hosted OpenProject, Snipe-IT asset inventory, and Strapi CMS on Docker behind Caddy and Cloudflare Tunnel; added SonarQube + DefectDojo for AppSec.
Lower hosting spend and one monitored, auditable stack, published through tunnels instead of open ports.
DockerCloudflare TunnelCaddyGrafanaSnipe-ITSonarQubeDefectDojo
Open Source
Open Source: Session Pulse for Claude Code
Developers running several AI coding sessions had no simple way to see what each was doing or where their time went.
A Python package with a local dashboard and tray indicator: live sessions, token use by model, active hours, a to-do board, and AI-generated follow-ups.
Published on PyPI under MIT, with CI on Python 3.11–3.13.
PythonPyPIGitHub ActionsOpen Source
Kubernetes & DR
AWS EKS Infra Optimization + DR
SaaS workloads needed reliable ingress, backups, and alerting on EKS.
Improved ingress, added Velero backups, tuned Prometheus/Alertmanager, GitLab CI pipelines.
Higher reliability and DR readiness with better observability.
EKSVeleroPrometheusAlertmanagerGitLab CI
CI/CD
Fastlane + GitHub Actions: 50+ iOS Apps to TestFlight
Manual deployment across 50+ apps was slow and error-prone.
End-to-end CI/CD with Fastlane, GitHub Actions, signed builds, lane conventions, reusable workflows.
Release time: hours → minutes. 100% consistency across apps.
GitHub ActionsFastlaneiOSCI/CD
Observability
Company-wide Central Monitoring & Logging
Fragmented observability across servers and services.
Prometheus + Grafana + Loki + Promtail with exporters and SLO-based alerting.
Downtime −40%, incident detection +70%.
PrometheusGrafanaLokiPromtail
Secrets Management
Multi-Cluster K8s Secret Management with Vault
Secret sprawl and inconsistent auth across clusters.
Vault with multi-cluster auth, JWT/Kubernetes methods, standardized secret engines.
Simplified secrets mgmt; audit-friendly; safer deploys.
VaultKubernetesSecurity
