Rahul K
Site Reliability Engineer @ Cedana (YC S23)
I specialize in DevOps, platform, systems and HPC engineering, I've helped companies scale to 10,000+ CPU and GPU workloads. I love tinkering with low-level Kubernetes and designing platforms in Go and Rust.
On the side, I run nixos as my daily driver with neovim and qtile TWM. I love playing bullet chess (rated around 2000). I read about handwriting analysis and philosophy when the time permits.
Experience
Remote
Cedana
Site Reliability Engineer
- Built Cedana-managed infrastructure for Caltech through cedana-cli, running 6,000+ CPU and 2,000+ GPU workloads on EKS.
- Redesigned plugin binary distribution in Rust to go through S3 via an API, saving $12,000+ a year. The change needed linked refactors across 6+ repos. I added CRC32 checksum verification and fixed CI network regressions with S3 signed URLs.
- Led the design of Slurm restore observability: a SPANK-spawned Go monitor tracks CRIU restores and reports profiling stages.
- Led the migration of the legacy Go backend to Rust, and a full refactor of the Kiota-generated propagator SDK.
- Owned the Rust policy engine that runs custom checkpoint/restore actions on Kubernetes and Slurm, with SQL-based throttling.
- Designed cgo SPANK hooks and signal traps for preemption-driven Slurm checkpoint/restore, plus the Ansible installer.
- Built a Kubernetes operator that streams Vector.dev metrics to ClickHouse. Hive pathing cut payloads from 4 MB to 24 KB and RAM use from 80% to 20%.
- Engineered an H100 end-to-end testing framework for Kubernetes checkpoint/restore across OCI container runtimes.
- Deployed EKS Karpenter through CloudFormation, with custom-AMI GPU node pools that scale to zero on weekends.
- Ran ML/LLM workloads (vLLM, SGLang, OpenCode, PyTorch, co-folding) on Kubernetes with live CUDA checkpoint/restore.
Bengaluru
CloudSEK
DevOps Engineer
- On call for core product releases. Built the infrastructure for applications that scan 3,000,000 websites.
- Maintained the ArgoCD GitOps workflow, Helm monorepo, KEDA and GitLab CI pipelines for prod and staging.
- Set up Grafana, Zabbix, Lambda and SigNoz alerts to Slack, and PagerDuty call alerts for Uptime Kuma.
- Implemented a Wazuh SIEM architecture for SOC 2 compliance. Migrated legacy EKS clusters to GKE with Velero.
Bengaluru
CloudSEK
DevOps Intern
- Set up a Kubernetes-based Qdrant installation for ML applications, and added Kubecost for cluster cost visibility.
Remote
Audio AI
DevOps Intern
- Built AWS infrastructure for MLOps, with Jenkins master-agent CD that ships models in Cog containers.
Remote
Muziriz Tech
Software Developer Intern
- Explored ways to trace HTTP/HTTPS with eBPF and worked on Aqua Security's Tracee, raising HTTPS coverage to 70%.
- Measured and optimized the performance of Logstash and Filebeat.
Skills
Projects
Kettle: container runtime engine
A minimal container runtime engine designed to emulate a containerd-like architecture: create, run and delete containers, pull remote images and keep overlayfs snapshots.
Browser Orchestrator
A Rust-based microservice orchestration system aided by Restate for spawning multiple browser sessions exposed through a single axum server.
eBPF Tools (hive)
hive is a CLI consisting of networking tools which can be used to trace/redirect network traffic.
Stingless
A monitoring system with flamegraphs, queue alerts and more.
Honors & community
- Reviewed 30+ talk proposals as a program reviewer for the FOSS India conference.Sep '26
- Two-time distinguished panel member and judge at the AEGIS cybersecurity hackathon.Mar '25 & Mar '26
- Recognized for entrepreneurial excellence at Entrepreneur Conclave, DSCE.Dec '24
- Technical lead at Pointblank Club, guiding and teaching junior members.Sep '22 – now
- Former team lead of the DSCE Chess Club. Organized online tournaments that reached 80+ students.Jul '23 – Jul '24
Education
Bengaluru
Dayananda Sagar College of Engineering
B.Tech, Computer Science · CGPA 9.0