Built and operated 5 production systems on multi-AZ AWS with full IaC + observability. Currently leading ElastiKube (ML-driven K8s autoscaling) and TesseraKt (managed K3s SaaS — VPC → SG → EC2 → k3s cluster in a single POST /api/v1/clusters).

⚡ 1.53ms     Go redirect P95 — 2–7× faster than Python baseline
💰 60%         AWS cost reduction via ML-driven K3s autoscaling
📈 99.82%      Cache hit rate · 1,000+ RPS · 0% errors (Snipl production)
🌐 Multi-AZ    Production-grade AWS deployments with IaC + observability
🖥️  11+        EC2 instances automated across 3 AZs
🔄 4-layer     Autoscaling architecture (time-aware · flash-sale · Prophet ML · reactive HPA)
📦 1K+         Concurrent users load-tested
📦 70+ OSS repos · 5 production systems shipped

60% AWS cost reduction through a 4-layer scaling architecture: time-aware policies, flash-sale detector, Prophet ML forecasting, and reactive HPA — coordinated via event-driven Lambdas and DynamoDB state.

• Multi-AZ round-robin worker placement, DynamoDB distributed locks, WAL crash recovery
• Spot instances with automatic On-Demand fallback; 17 CloudWatch alarms
• Stack: AWS Lambda · EventBridge · DynamoDB · EC2 · Prometheus · K3s · CloudWatch

Polyglot microservices with a Go Chi redirect service hitting P95 1.53ms / 99.82% cache hit / 1,000+ RPS — 2–7× faster than the Python baseline. Live on Render + Cloudflare.

• Hybrid Go + Python stack (Go Asynq workers alt to Celery); Redis cache-aside, MongoDB projection (~60% payload cut)
• PostgreSQL FOR UPDATE SKIP LOCKED advisory locks; PgBouncer pool; full OTel + Tempo/Loki/Grafana
• Stack: Go · Python · FastAPI · Chi · Asynq · Redis · PostgreSQL · MongoDB · OTel · Pulumi · Ansible

Django REST + Celery pipeline transcoding 480p → 4K DASH with ~70% size reduction. Token-secured media via Nginx X-Accel-Redirect + tiered throttling (anon/auth/upload/stream).

• End-to-end CI/CD: GitHub Actions → Pulumi → Ansible
• Stack: Django · Celery · FFmpeg · AWS S3 · Nginx · Pulumi

Email/SMS/Push via SendGrid, Twilio, FCM. Direct pika consumer (no Celery), idempotent worker, exponential backoff (1s/2s/4s).

• JWT scoped service-to-service auth; Redis token-bucket rate limiting (100 req/min + 20 burst); 30s cache-aside
• Stack: FastAPI · RabbitMQ · pika · PostgreSQL · Redis · JWT

Four-service split (API · Worker · Monitor · Redis) with priority queues, dependency DAG, auto-scaling by load-per-worker via Docker Swarm, dead-letter queue, and a real-time web dashboard.

• Interactive stress harness (light / medium / heavy / dependency / priority / custom)
• Stack: FastAPI · Redis · Docker Swarm · Jinja2

Customer + admin dashboards with OTP-verified auth, dynamic installment calculator, and weekly/monthly/all-time financial reports.

• Stack: React · Vite · TypeScript · Tailwind · shadcn/ui · FastAPI · PostgreSQL · Redis · Celery · Alembic

PostgreSQL replication with automated backup/recovery, Redis Sentinel for HA caching, Nginx LB, full Pulumi IaC across two AZs.

TesseraKt — managed Kubernetes SaaS in Go + Python. Provisioning pipeline ships end-to-end against real AWS: VPC → multi-AZ subnets → security groups → EC2 → k3s cluster (control plane + workers) from a single POST /api/v1/clusters. Cluster deletion, autoscaling, and in-cluster observability bootstrap are the next milestones.

Full-Stack Developer (Contract) — Independent Practice (Mar 2026 – Present, Remote) Solo contractor building a CRM for an Australian NDIS provider — participant onboarding wizard, NDIS plan/funding tracking, staff roster, timesheets, e-signature capture, PDF generation. I own architecture, data model, UX direction, and backend; coding agent is used only for routine UI scaffolding, which I review and own end-to-end.

• Backend: Django 5.2 + DRF split across 8 apps (users · roles · teams · staff · roster · timesheets · participants · core) — custom RolePermission with permission substitutions, {data, error} response envelope, camelCase wire, services pattern with transaction.atomic + select_for_update, integer PKs for system / UUID PKs for domain models
• Frontend: Vite + React 18 + TS + Tailwind + shadcn/ui + Radix + TanStack Query — feature folders (form-builder · roster · teams · permissions) Stack: Django 5.2 · DRF · SQLite → PostgreSQL · Vite · React 18 · TypeScript · Tailwind · shadcn/ui · TanStack Query.

Backend Engineer & Product Builder (Remote, Aug 2024 – Present) Designing and operating 5 production-grade backend systems across messaging, async pipelines, video, and infrastructure. Acting solo architect/operator: IaC, deployment, observability, incident response.

• Shipped systems across multi-AZ AWS with full IaC + observability stack
• Replaced manual provisioning with one-command Pulumi + Ansible deploys
• Built ElastiKube K8s autoscaler: 60% AWS cost reduction with ML predictive scaling

Backend Developer @ Cooking Station (Jun 2024 – Aug 2024, Dhaka) Role-based admin dashboard for 200+ users, automating 40% of manual account workflows. Cron-based scheduling system. Stack: Django · MySQL · Docker.

• B.Sc. Computer Science & Engineering — Daffodil International University (2017–2022)
• System Design & Backend Engineering with AWS — Poridhi (Intensive production engineering: 100+ hands-on labs in K8s, AWS, distributed systems)

• 📝 Building a Scalable URL Shortener: System Design to Production
• 📺 I.T. Darshonik YouTube — 40+ Bengali tech videos
• 🌍 Quora Bengali tech blog — 200K+ readers

Looking for an engineer who can own infrastructure end-to-end, ship polyglot microservices, and automate from commit to production?

📧 Email: kaziiriad@gmail.com 💼 LinkedIn: Sultan Mahmud 📝 Blog: Medium 📺 YouTube: I.T. Darshonik