// Senior DevOps Engineer · AI Agent Builder · Bengaluru
7+ years building cloud-native infrastructure and automating complex CI/CD pipelines — slashing deployment cycles from 3–4 days to 2 hours with Kubernetes, Terraform and Jenkins. These days I also build autonomous AI agent systems — multi-agent orchestration with LangChain, LangGraph and local LLMs.
I'm Anjul — a Senior DevOps Engineer at Oracle, Bengaluru, where I lead automation and infrastructure for core telecom products. My job, in one line: make deploys boring. That means infrastructure as code, pipelines that catch problems before production does, and observability that surfaces issues before users notice.
I've worked across OCI, Google Cloud, AWS and OpenStack — from migrating 200+ servers to GCP, to running Kubernetes upgrade cycles across live demo environments with zero data loss, to cutting cloud spend by re-architecting where workloads run. I also mentor engineers and push DevOps best practices across teams, because good infrastructure is a team sport.
Beyond infrastructure, I build autonomous AI agentic systems — designing multi-agent architectures with LangChain and LangGraph, RAG pipelines with vector databases, and fully local LLM deployments. My flagship is Brahmaand — a 12-agent autonomous system that runs entirely on local models at zero cloud cost.
$ whoami → senior devops engineer · ai agent builder · mentor
Sep 2024 — Present
Sep 2022 — Aug 2024
Oct 2020 — Aug 2022
Nov 2018 — Oct 2020
Owned and executed the K8s upgrade cycle (to v1.36) across three live demo environments — planned windows, sequenced rollouts, zero data loss, all apps validated healthy post-upgrade.
Re-architected demo clusters from managed cloud Kubernetes to on-prem kubeadm on OpenStack — eliminating per-node compute, load balancer, storage and egress charges per environment.
Drove cloud security remediation — bastion migration to private subnets, least-privilege access enforcement, and tightened deletion controls protecting critical environment data.
Delivered customer-facing demo environments end to end — auto-scaling and rolling upgrade/rollback use cases, certificate and pipeline coordination, and production-blocking image fixes.
⭐ Flagship AI Project
autonomous multi-agent AI system · Sanskrit for "Universe"
A fully autonomous, containerized multi-agent AI system I designed and built from scratch — 12+ cooperating agents modeled on Hindu cosmology, where each deity's cosmic role maps to its function in the system.
# one message → the right agent acts You (Telegram) │ ▼ [ INDRA ] ─ gateway │ ▼ [ PARAMATMA ] ─ supreme router │ ┌────┼─────────┐ ▼ ▼ ▼ BRAHMA VISHNU SHIVA create preserve destroy │ │ │ ▼ ▼ ▼ agni krishna rudra moon rama nataraja narasimha mahakala │ ▼ result → Telegram ✓
Free, open-source resume builder — 20 templates, upload & auto-parse your existing resume, ATS keyword checker, clean PDF export. 100% in your browser — nothing is ever uploaded. Built so students and professionals can make a great resume without paying anyone.
Try it — resume.cloudx.co.in →Autonomous multi-agent AI system — 12+ cooperating agents with hierarchical routing, a Redis message bus and a Telegram gateway, all on local LLMs at $0/month cloud cost.
Read the full breakdown ↑Full-stack NIFTY/BankNifty options cockpit + autonomous signal engine — an 11-indicator weighted confluence score (EMA, Supertrend, VWAP, MACD, RSI, PCR…) drives entries with ATR stop-losses, 1:2 targets, trailing SLs and auto EOD flatten. Currently in paper-testing before going live.
Private — results after paper-testing wrapsThis very site — hand-coded vanilla HTML/CSS/JS with zero frameworks and zero build step, deployed to GitHub Pages by a plain git push. Bonus: an immersive Three.js WebGL experience with a glass-crystal hero, GLSL shaders and 3D-tilt cards lives at /lab.html.
View source on GitHub →Automation framework for Oracle's core telecom products, built from the ground up — took release deployments from 3–4 days of manual work to 2 hours, with observability wired in via Prometheus, Grafana and Loki.
Proprietary — happy to talk through the designReplaced managed cloud Kubernetes (OKE) with kubeadm clusters on OpenStack for demo environments — same K8s capabilities, zero per-node compute, LB and egress charges, with Longhorn/NFS for storage.
Proprietary — architecture in Highlights aboveHave a project, a role, or just want to talk cloud and Kubernetes? My inbox is always open — I'll get back to you fast.
Say Hello 👋