Open to FDE, AI Platform & Cloud Platform roles

Hi, I'm Akhilesh

Cloud, Platform & DevSecOps Engineer heading toward Forward Deployed Engineering

Building production systems at the intersection of Cloud, Kubernetes & AI.

7+ years of experience building, automating and operating cloud platforms. I combine production engineering, infrastructure automation and reliability with growing hands-on work in AI platforms and intelligent systems.

  • GCP
  • Kubernetes
  • Terraform
  • Python
  • APIs
  • GenAI
Illustrated portrait of Akhilesh Ranjan Singh surrounded by cloud, Kubernetes and AI sketches

About

A production engineer, extending into AI

Photo of Akhilesh Ranjan Singh

I'm a Cloud, Platform & DevSecOps Engineer with 7+ years of experience building and operating production systems across GCP, Azure and AWS.

My work spans Cloud Platform Engineering, Kubernetes, Terraform, CI/CD, DevSecOps, cloud security and SRE. I'm now applying that production-engineering background to AI platforms, MLOps, RAG and Agentic AI systems.

I'm particularly interested in Forward Deployed Engineering — working closely with users, understanding ambiguous problems, rapidly building practical solutions and taking them all the way to production.

  • BuildProduction systems
  • SolveAmbiguous problems
  • DeployCloud & AI workloads
  • ImpactReal-world outcomes

Engineering approach

What I Bring to Forward Deployed Engineering

The way I approach platform engineering — applied end to end, from an ambiguous problem to a reliable system running in production.

  1. 01

    Discover

    Understand the problem

    Understand user and business problems, requirements and technical constraints.

  2. 02

    Build

    Prototype with Python/APIs/AI

    Prototype practical solutions using Python, APIs, cloud and AI.

  3. 03

    Deploy

    Productionize with Cloud/Kubernetes

    Productionize applications using Kubernetes, Terraform, CI/CD and GitOps.

  4. 04

    Operate

    Observe & improve

    Observe, troubleshoot, secure and continuously improve production systems.

The point of every stage: impact for the people using the system.

Experience

7+ years of production engineering

  1. Senior DevSecOps Engineer

    Bupa Capability Centre (BCC)

    Current – Present

    Security · Platform · DevSecOps

    Working within the Cyber Security Platform / CISO engineering function, focused on secure cloud platforms, automated delivery and DevSecOps engineering.

    • Engineering secure CI/CD pipelines using Azure DevOps, integrating security controls throughout the software delivery lifecycle.
    • Building and managing cloud infrastructure using Terraform / Infrastructure as Code, with security and governance embedded by design.
    • Integrating SAST, DAST and SCA into delivery pipelines for earlier vulnerability detection and remediation.
    • Working with Docker, Kubernetes, FluxCD, Helm and Kustomize to support cloud-native and GitOps-based delivery.
    • Implementing OPA / Policy as Code, infrastructure hardening, secure configuration and secrets management with Azure Key Vault.
    • Automating platform and security workflows using Python, PowerShell and Shell scripting, while working across Azure, PaaS services, APIs, monitoring and logging.
    • Azure
    • Azure DevOps
    • Terraform
    • Kubernetes
    • Docker
    • FluxCD
    • Helm
    • Kustomize
    • OPA
    • Python
    • PowerShell
    • Azure Key Vault
  2. DevOps Engineer

    Capgemini · Noida

    –

    GCP · Kubernetes · Terraform · SRE

    • 70%less infrastructure provisioning effort
    • 20%cloud cost reduction
    • Designed and governed enterprise GCP landing zones — Shared VPC, project hierarchies, IAM governance and policy-as-code — for secure, compliant cloud adoption.
    • Built reusable Terraform Enterprise modules for networking, IAM, GKE, monitoring, security and AI infrastructure, cutting provisioning effort by 70%.
    • Enabled self-service infrastructure by integrating Terraform Enterprise with Git-based CI/CD, Harness, Jira and ServiceNow, streamlining change management.
    • Operated production GKE — cluster and node-pool lifecycle, upgrades, autoscaling, RBAC — and drove SRE practice through monitoring, incident response, RCA and capacity planning.
    • Reduced cloud spend by 20% through rightsizing, autoscaling, capacity planning and governance.
    • Designed and implemented MLOps workflows on GCP using Vertex AI Pipelines, Model Registry, MLflow and FastAPI, alongside GenAI/agentic workflows using Gemini, RAG and LangGraph.
    View more
    • Built CI/CD and GitOps workflows with Jenkins, GitHub Actions, Harness, ArgoCD, Docker, Helm and Terraform across multiple environments.
    • Hardened edge security and observability with Cloud Armor WAF policies, rate limiting, Adaptive Protection, OpenTelemetry, and attack-surface monitoring using IONIX and Stream Security.
    • Led a GitHub-to-Bitbucket migration and repository standardization, improving source-code governance and CI/CD integration.
  3. Senior Software Engineer – DevOps

    Tech Mahindra · Mumbai

    –

    Multi-cloud · CI/CD · Kubernetes

    • 99.9%service availability maintained
    • Drove CI/CD platform engineering with Jenkins, GitHub Actions, Terraform, Docker, Kubernetes, Helm and Ansible, improving release efficiency and deployment reliability.
    • Delivered Infrastructure as Code across GCP, AWS and Azure with Terraform and Ansible for provisioning, configuration management, backup automation and disaster recovery.
    • Ran highly available Kubernetes platforms with Docker and Helm, maintaining 99.9% service availability through proactive monitoring.
    • Implemented monitoring and centralized logging with Prometheus, Grafana and the ELK Stack for faster incident troubleshooting.
    View more
    • Strengthened platform security with DevSecOps controls and secure artifact management practices.
    • Managed backup, disaster recovery and data protection with Veeam and Kasten K10 for cloud-native workloads.
    • Integrated S3, GCS and Azure Blob storage with CI/CD pipelines and Kubernetes workloads for artifacts, state files and application data.
  4. System Engineer – Cloud & DevOps

    TCS · Bengaluru

    –

    Cloud · Automation · DevOps

    • 40%less deployment effort
    • Automated CI/CD with Jenkins, Cloud Build and Terraform, reducing deployment effort by 40% across GCP environments.
    • Managed production Kubernetes and GKE environments for business-critical applications.
    • Supported GCP provisioning, cloud migration and reliability work through Terraform automation, IAM governance and Linux administration.
    • Owned monitoring, alerting, incident management and RCA with application, infrastructure and security teams to maintain SLA compliance.

Projects

Things I've built end to end

Public reference implementations. Each repository documents what runs locally and what is written for your own cloud account.

FDE · Simulated customer engagement

Customer Self-Service Deployment Platform

Problem
A fictional customer, Meridian Labs: developers waited 2–3 days and filed several tickets across DevOps, infrastructure, Kubernetes and security to ship one microservice.
Solution
A self-service API and form that render the repository, CI pipeline, Terraform, Helm values, Kyverno policies and an Argo CD Application for dev or staging in under two minutes.
Constraint
The customer's last incident was a laptop apply, so the API stops at files. Argo CD is the only writer to a cluster, and prod is a pull request, not an API call.
Key decisions & outcome
Key decisions
  • prod requests are stored and returned as HTTP 422; the same idempotency key returns the original decision.
  • A local policy checker rejects images tagged latest and pods without limits before anything reaches Git.
  • Terraform, Helm and policy checks run on a laptop with no cloud login; GKE is documented, not required.
  • 36 issues across five milestones, each naming the customer impact and the file that closed it.
Outcome
A runnable engagement artifact with tests, a demo script and Prometheus/Grafana metrics. The customer is fictional; the engineering runs locally.
  • FastAPI
  • Terraform
  • Helm
  • Argo CD
  • Kyverno
  • Prometheus

Platform · IaC reference

Terraform GCP Landing Zone

Problem
Hand-built projects lead to snowflake networks, over-privileged service accounts, public clusters and scattered state.
Solution
Reusable Terraform modules (the "how") composed per environment — dev, staging, production (the "with what").
Architecture
Per environment: Shared VPC, Cloud Router + NAT, private GKE, Artifact Registry, Secret Manager, least-privilege IAM with Workload Identity, Cloud Armor and alerting, with state in GCS under isolated prefixes.
Key decisions & outcome
Key decisions
  • Private nodes with no public IPs; egress only via Cloud NAT; deny-by-default firewall.
  • Per-environment state isolation, so a dev apply can never plan changes to prod.
  • Environment sizing: spot nodes and preview-mode WAF in dev, a 3-zone spread and enforced Cloud Armor in prod.
  • Checkov and OPA/Conftest policies plus keyless OIDC in GitHub Actions CI.
Outcome
An infrastructure-as-code reference: terraform validate and fmt clean, not applied to a live account.
  • Terraform
  • GCP
  • GKE
  • IAM
  • Cloud Armor
  • OPA

Delivery · GitOps reference

GitOps Kubernetes Deployment

Problem
Manual kubectl apply causes drift, no audit trail, risky all-at-once releases and diverging environments.
What I built
A FastAPI service delivered dev → staging → production through PR-based promotion with Argo CD.
How it works
Kustomize overlays per environment; Argo Rollouts canary in staging and blue-green in production, gated by Prometheus analysis and Kyverno policies.
  • Argo CD
  • Argo Rollouts
  • Kustomize
  • Kyverno
  • FastAPI
  • Prometheus

CI/CD · Reusable library

GitHub Actions DevOps Templates

Problem
Every repo copying its own pipeline YAML means drift, uneven security scanning and long-lived cloud keys.
What I built
A library of reusable workflows for Python, Node and Terraform CI, container builds, security scanning, GKE deploys and releases.
How it works
Consumers call versioned workflow_call pipelines; OIDC/Workload Identity for cloud auth and automatic rollback on failed deploys.
  • GitHub Actions
  • OIDC
  • Trivy
  • Docker
  • GKE
View more projects

Algorithms · Kubernetes concepts

Rolling Update Simulator

A dependency-free Python rollout engine that moves a fleet from v1 to v2 without exceeding resource limits or dropping below minimum availability — and rolls back safely when the new version keeps failing health checks.

  • Python
  • FastAPI
  • Locust
  • Kubernetes

FDE / AI Engineering Lab

Simulated customer engagements, learning ladders and coursework — labelled as such. The customers and data are fictional; the code, tests and constraints are real and run locally.

Simulated engagement · Logistics

Harborline Dispatch Copilot

Shipment risk score with every point explained, the approved SOP and cited sources. Shipment data never leaves the environment and no external model is called; the audit log keeps the decision, not the ticket text.

  • Python
  • FastAPI
  • Eval gate
  • Docker

Simulated engagement · Healthcare ops

Northshore After-Hours Intake

Routes overnight portal notes for the morning nurse and blocks any note without consent. It routes and cites; it does not diagnose, and note text is never sent to an outside model.

  • Python
  • Consent gate
  • Citations

Simulated engagement · Payments

Brightpath Payment Reconciliation

Idempotent matching of webhooks to the settlement file. Duplicate events are ignored, and any cent mismatch goes to an exception queue instead of being marked settled.

  • Python
  • Idempotency
  • Exception queue

Simulated engagement · Field service

Helios Field Parts

Offline spare-part lookup for a depot with no reliable uplink. It picks the bin that actually has stock, reports stockouts, and escalates unknown faults instead of guessing. It never creates a purchase order.

  • Python
  • Offline-first
  • Inventory rules

Simulated engagement · Utilities

Cedar Grid Outage Desk

Ranks storm outages so a feeder with a life-safety account goes ahead of a larger commercial outage, and suggests crew counts. It does not text customers or dispatch crews on its own.

  • Python
  • Prioritisation
  • Human in the loop

Simulated engagement · Procurement

Northline Vendor Review

Checks vendor packets for expired insurance and bank-detail changes, and enforces that the person who entered a change cannot accept it. Software can mark a packet ready for a human, never approved.

  • Python
  • Four-eyes control
  • Audit

Method · Templates

FDE Engagement Playbook

How I run an engagement: discovery brief, success metrics agreed before code, solution design, security and data rules, a shadow-week rollout and the customer readout. The engagements above are built from these templates.

  • Discovery
  • Success metrics
  • Rollout

Learning ladder · Platform

Platform Engineering Ladder

Small tested FastAPI services that climb from host health, cost and log analysis through Terraform and Kubernetes rendering to multi-tenant auth and a self-service developer platform. Each renders or plans; none applies to production.

  • FastAPI
  • Terraform
  • Kubernetes
  • pytest

Learning ladder · Data & AI

Selected Skill Ladder

A progression from Python fundamentals and data science through ML, NLP and RAG — PDF question answering, BM25 hybrid search and a RAG evaluation gate — implemented without hosted models so every score is reproducible.

  • Python
  • ML
  • RAG
  • Evaluation

Learning ladder · FDE progression

20-Project AI + Platform Path

Public labs from host health and analytics through churn, fraud, production ML, RAG eval, MCP CloudOps, an internal developer platform, and an agentic ops capstone. Each is a local proof: extractive retrieval, tests, and no production apply.

  • FastAPI
  • ML
  • RAG
  • MCP
  • GKE labs

Coursework · MLOps

MLOps Zoomcamp

Hands-on exercises for productionizing ML services: experiment tracking, orchestration, deployment and monitoring.

  • Python
  • Jupyter
  • MLflow

Skills

Technical capabilities

Grouped by what I use them for. No self-rated percentages — just what I rely on versus what I've worked with.

Core daily drivers Experienced used in delivery work

Cloud & Platform

  • GCP
  • Kubernetes
  • Terraform
  • Docker
  • GKE
  • Helm
  • Azure
  • AWS
  • Ansible

Delivery & Reliability

  • CI/CD
  • GitOps
  • Observability
  • SRE
  • Jenkins
  • GitHub Actions
  • ArgoCD
  • Harness
  • Cloud Build
  • Azure DevOps
  • FluxCD
  • Kustomize
  • Prometheus
  • Grafana
  • OpenTelemetry
  • ELK

Programming & Automation

  • Python
  • REST APIs
  • SQL
  • Bash
  • FastAPI
  • Go
  • Java

AI Engineering

  • RAG
  • LLM APIs
  • LangGraph
  • Vertex AI
  • Gemini
  • LangChain
  • MLflow
  • Kubeflow
  • Ollama

Security

  • DevSecOps
  • IAM / RBAC
  • SAST / DAST / SCA
  • OPA / Policy as Code
  • Azure Key Vault
  • Workload Identity
  • Cloud Armor / WAF
  • Prisma Cloud

Data

  • PostgreSQL
  • MySQL
  • MongoDB
  • Vector databases

Credentials

Certifications, practice & education

Cloud certifications

Engineering practice

Continuous learning

Education

Bachelor of Engineering – Computer Science RGPV, Bhopal

2019CGPA 8.55

Contact

Let's Build Something

I'm exploring opportunities across Forward Deployed Engineering, AI Platforms and Cloud/Platform Engineering.

akhileshranjan.ks@gmail.com · Noida, India

Send a note

Opens your email app with the message filled in.