Senior DevOps & AI Engineer · CKA · AWS SAA-C03
Bangkok, Thailand · [email protected]
⬇ Download Resume (PDF)
I started my career managing 50+ Linux servers in Nepal for an organisation bringing education to kids in remote villages with no reliable internet. Back then, “infrastructure automation” meant a bash script that maybe worked. Today I’m building AI agents that autonomously manage cloud infrastructure and talk to AWS like it’s a conversation.
That arc — from hand-configured servers in Kathmandu to multi-agent LangChain systems on AWS Bedrock in Bangkok — is basically my entire personality.
I’m Bikram Dhoju, a Senior DevOps & AI Engineer based in Bangkok, Thailand. I’ve spent 10+ years making sure things don’t fall over at 2am — and increasingly, building systems smart enough to fix themselves before anyone gets paged.
My obsessions, in rough order:
- Kubernetes — I find it genuinely satisfying when a cluster self-heals. I’ve led migrations from monoliths that took 45 minutes to deploy to microservice platforms that ship 300% more frequently.
- Infrastructure as Code — if it’s not in git, it doesn’t exist. Terraform, Terragrunt, and ArgoCD are how I sleep at night.
- AI agents for ops — I built a multi-agent platform using LangChain and AWS Bedrock that handles infrastructure tasks through natural language. It reduced manual operations by 70%. The on-call rotation noticed.
- Observability — I have strong opinions about dashboards. I run the full LGTM stack: OpenTelemetry Collector → Loki + Tempo + Prometheus → Grafana, with trace-log correlation so you can jump from a slow span straight to the log line that explains it. A good alert rule is worth more than three engineers staring at logs.
- Security — I automated CloudTrail monitoring into a system called Sentinel because knowing about the security event two hours later is not knowing about it.
Things I’ve built that I’m proud of
Cut AWS infrastructure costs by 20% with intelligent autoscaling on custom business metrics — not CPU, actual business signals. Deployed an enterprise RAG system over Confluence and runbooks that cut incident resolution time by 40% because the answer was always in the docs, nobody could find it. Built Kubernetes admission webhooks in Go that enforce compliance policy at deploy time rather than audit time. Rolled out a full LGTM observability stack — OpenTelemetry Collector feeding traces into Tempo, logs into Loki, metrics into Prometheus, all correlated in Grafana — cutting incident detection to under 5 minutes.
The longer version
2015 → Linux sysadmin, Nepal. 50+ servers. Ansible before it was cool.
2020 → SRE at an ML company. 1M+ predictions/day. Learned what uptime really costs.
2021 → Senior DevOps, Bangkok. Kubernetes at scale. GitOps everywhere.
2022 → Lead engineer. Multi-account AWS. Custom K8s webhooks in Go.
2023 → AI Engineer. LangChain. AWS Bedrock. Autonomous infra.
Now → Still figuring out how to make it all break less.
Technical toolkit
Cloud — AWS (EKS, ECS, RDS, Lambda, Bedrock, Cognito) · Azure (AKS) · GCP (GKE, BigQuery)
AI & Agents — LangChain · Anthropic API · OpenAI · RAG · pgvector · MCP · Multi-agent orchestration
Platform — Kubernetes · Helm · Kustomize · Istio · Linkerd · KEDA · Docker
IaC — Terraform · Terragrunt · Ansible · CloudFormation
CI/CD — ArgoCD · FluxCD · GitHub Actions · Flagger · Canary deployments
Observability — OpenTelemetry (OTEL) · Prometheus · Grafana · Loki · Tempo · DataDog · ELK · Graylog · Distributed tracing
Code — Python · Go · Bash (and enough JavaScript to be dangerous)
Security — HashiCorp Vault · GuardDuty · CIS Benchmarks · IAM · Cloudflare
Certifications
Certified Kubernetes Administrator (CKA) · AWS Solutions Architect – Associate (SAA-C03)
Background
Electronics & Communication Engineering, Pulchowk Campus, Tribhuvan University (2014). Specialised in IPv6 networking and remote sensing — which turned out to be surprisingly relevant to building distributed systems that span multiple regions.
Find me on GitHub or reach out at [email protected]. Bangkok, Thailand.