ANKIT KUMAR
PANDITAI & Cloud Infrastructure at Scale
I architect cloud infrastructure that scales, Kubernetes platforms that self-heal, and CI/CD pipelines that make deploying feel like breathing. Based in Nepal, building for the world.
Add your photo topublic/images/photo.jpg
Not just an engineer —
a craftsman
Results-driven DevOps Engineer and Infrastructure Architect with 5+ years of experience owning end-to-end AWS infrastructure for Web3 and AI SaaS platforms. Specialized in Kubernetes at scale, serverless architecture, FinOps, and GitOps-driven infrastructure management using Terraform and GitHub Actions.
Hands-on expertise across the full Web3 stack — RPC/blockchain nodes, indexers, subgraphs, wallet/key infrastructure, and smart contract deployments — now extended to GPU and inference infrastructure for AI workloads. My focus is on reducing infrastructure costs, eliminating manual interventions, and building reliable, auditable systems that scale confidently across multi-tenant environments.
Alongside full-time engineering, I run an independent practice delivering web application development, cloud architecture, and FinOps engagements for teams that need senior infrastructure depth without a full-time hire.
I believe the best infrastructure is invisible—it just works. When systems are truly well-built, engineers ship faster, and your team sleeps better. That's my goal.
Tools I own
completely
Where I've
shipped
Owning system design and the cloud/GPU infrastructure foundation for an early-stage AI product — translating product requirements into architecture, then taking it from zero-to-one through training and inference workloads running reliably in production.
- Led end-to-end system design — service boundaries, data flow, storage and queueing choices, and failure/scaling characteristics documented before a line of infrastructure was written
- Designed the initial AWS landing zone from scratch — multi-account structure, VPC networking, IAM boundaries, and Terraform state management
- Built GPU-backed Kubernetes node pools with autoscaling and spot capacity for training and inference workloads
- Shipped GitOps CI/CD for model and application releases, giving the team reproducible, one-click deploys
- Instrumented cost visibility from day one — per-workload tagging and budget alerts to keep GPU spend predictable while scaling
- Set up observability and on-call foundations covering latency, GPU utilisation, and inference error rates
Independent engagements across web application development, cloud architecture, and FinOps — helping startups and small teams get production-grade infrastructure without a full-time platform hire.
- Web application development — built and shipped full-stack apps on Next.js, Node.js, and serverless APIs, from first commit to live domain
- Cloud services — architected, migrated, and hardened AWS and GCP environments with Terraform, covering networking, IAM, backups, and disaster recovery
- FinOps — ran cost audits delivering rightsizing, spot adoption, storage lifecycle policies, and per-team cost allocation with measurable monthly savings
- Platform & CI/CD setup — containerised legacy workloads and delivered automated pipelines with staging/production parity
- Advised founding teams on cloud spend strategy, architecture trade-offs, and scale-readiness ahead of funding milestones
Sole DevOps owner driving infrastructure roadmap, incident response, capacity planning, and cross-team collaboration across a dual-purpose Web3 SaaS platform.
- Managed Kubernetes clusters at scale → achieving 99.9%+ uptime
- Architected and scaled multi-tenant AWS infrastructure with strict namespace and network isolation across dev, staging, production
- Deployed full-stack Web3 applications using blue/green and canary release strategies, eliminating downtime
- Established full-stack observability across Kubernetes, blockchain nodes, serverless functions using Datadog
- Drove measurable infrastructure cost reduction through right-sizing, spot instances, and automated cleanup
Designed and implemented fully automated CI/CD pipelines and AWS infrastructure for Web3 products, enabling reliable workload management at scale.
- Designed and implemented fully automated CI/CD pipelines using Jenkins and GitHub Actions → reducing deployment time by ~60%
- Provisioned and managed cloud infrastructure on AWS using Terraform, ensuring consistent, repeatable environments
- Managed container orchestration using Docker and AWS ECS, enabling reliable workload management at scale
- Automated configuration management with Ansible, reducing manual server setup and configuration drift
Developed and deployed smart contracts for DeFi products. Built comprehensive QA testing frameworks and contributed to deep understanding of on-chain mechanics.
- Developed and deployed smart contracts using Solidity and Python for DeFi products
- Built a QA testing framework in Jupyter Notebook, improving smart contract test coverage and reliability
- Gained deep understanding of on-chain mechanics and Web3 architecture
Things I've
built
A reusable Terraform module library (VPC, EKS, IAM, RDS) plus the GitHub Actions workflow that gates it — automated plan posted to the pull request for peer review, apply on merge, and remote state with locking. Replaced ad-hoc console changes with a reviewable change history.
Blue/green and canary release tooling for blockchain-facing services, where a bad deploy means dropped RPC traffic rather than a retryable request. Health-gated traffic shifting, automated rollback on error-rate regression, and per-namespace isolation across dev, staging, and production.
A repeatable cost-audit playbook applied across Web3 SaaS platforms and freelance client accounts — right-sizing EC2 and GPU instances, spot adoption, storage lifecycle policies, automated resource cleanup, and per-team cost allocation with anomaly alerting.
Built from scratch with Next.js 16, TypeScript, and CSS Modules — no UI framework, no template. Statically prerendered, responsive across four breakpoints, and driven entirely from a single typed content file. Source is public if you want to read it.
Validated
expertise
Start a
conversation
Whether it's a full-time role, a freelance engagement, a cloud cost audit, or just a technical discussion — I'm always happy to connect with people who care about building things well.