Hi there, I'm

ManmohanRawat

Solution Architect · AI Enthusiast · TOGAF® · Cloud Native Expert

“The models got smarter. Someone still has to keep the lights on.”

View ExperienceGet in Touch

Hi, I'm Manu! I design cloud platforms. And yes, I'm watching your cursor.

1 of 14
Current
Manager · MHP – A Porsche Company · since Mar 2023
Experience
13+ years
Location
Frankfurt, Germany

01 — whoami

About Me

Solution architect with 13+ years of designing, building and operating cloud-native platforms for financial services and automotive, from middleware and Pivotal Cloud Foundry to Kubernetes and the public cloud. Core stack: Kubernetes, Docker, AWS, Azure, Terraform, Datadog, Prometheus and TypeScript. Alongside client work, I build with generative AI hands-on: a serverless RAG library on AWS (Amazon Bedrock, DynamoDB) with tenant isolation, spend limits and rate limiting, and AI agents connected to real systems through the Model Context Protocol (MCP) for code review, migrations and infrastructure changes. My focus is the architecture around the model: retrieval quality, evaluation, least-privilege access, cost and observability.

Friends call me Manu.

Manmohan v13.0 — Changelog

Added:
telc Deutsch B1 certificate
Fixed:
fear of AI taking over jobs (partially)
Known issue:
German stuck at B1. Can handle a standup, can't win an argument in retro.
Deprecated:
Cloud Foundry. We had good times.

kubectl describe human manmohan-rawat

Name:
manmohan-rawat
Namespace:
frankfurt
Labels:
role=solution-architect, coffee=required
Status:
Running (uptime: 13y)
Restarts:
0 (a few near-misses on Fridays)
Limits:
remembers every outage, forgets every birthday
Events:
Warning BackOff der/die/das resolution failed, retrying...
location
Frankfurt, Germany
email
manmohan.rawat1391@gmail.com
status
Not actively looking for new opportunities

Languages

  • EnglishFull professionalfluent in "per my last email"
  • HindiNative or bilingual
  • GermanLimited workingtelc B1enough to survive a standup, not enough to run the retro
  • TamilLimited workingenough to know when you're talking about me
  • YAMLFluent, against my will

02 — rollout history

Experience

  1. current – Present

    Manager

    MHP – A Porsche Company · Frankfurt, Germany

    Cloud architect and AI engineer building shop-floor products for the logistics and automotive sectors.

    Currently at a Porsche company. No, I don't get a discount. Yes, I've asked.

  2. –

    Solutions Architect & Product Owner

    Wipro Limited · Düsseldorf, Germany

    Designed cloud-native solutions and owned product roadmaps across enterprise clients in Germany.

  3. –

    DevOps Engineer

    anynines · Germany

    Worked on cloud platform engineering and DevOps tooling for cloud-native workloads.

  4. –

    Team Lead – Cloud Foundry Operations

    Wells Fargo · Bengaluru, India

    Led the India cloud team managing PCF, PKS, Redis, and Concourse. Mentored teams migrating from legacy to cloud. Collaborated with Pivotal team in North Carolina on the Kubernetes (PKS) roadmap.

    • Led cloud team across regions on PCF, PKS, Redis & Concourse
    • Mentored teams in legacy-to-cloud migration
    • Collaborated with Pivotal team (NC) on PKS roadmap
  5. –

    PCF Operations Analyst → WebSphere Admin

    Ford Motor Company · Chennai, India / Detroit, USA

    Supported PCF on VMware & Azure across Sandbox, Preprod, and Production. Configured product tiles (RabbitMQ, Redis, MySQL, Spring Cloud). Integrated Splunk for platform monitoring. Earlier role as WebSphere Admin managing WAS/JBoss 4.x/5.x in Production, Dev, and QA.

    • PCF on VMware & Azure — Sandbox → Production
    • Tiles: RabbitMQ, Redis, MySQL, Spring Cloud, SSO
    • Splunk integration for platform health monitoring
    • WAS/JBoss admin with ITIL incident & change management
  6. –

    Senior System Administrator

    Cognizant · India

    Associate Integrator on the Middleware Integration Services Team. Delivered optimized middleware deployment solutions and provided 2nd-level production support.

03 — service mesh

Skills & Stack

Cloud & Infra

  • Kubernetes – certified
  • Cloud Foundry
  • PKS
  • Docker
  • Microsoft Azure – certified
  • AWS – certified
  • IBM WebSphere

IaC & Observability

  • Terraform – certified
  • Prometheus
  • Datadog
  • Concourse
  • Splunk

Languages & Dev

  • Typescript
  • Golang

Architecture & Leadership

  • TOGAF® – certified
  • Solution Architecture
  • Product Owner – certified
  • DevOps
  • Project Management
  • ITIL

Backed by a certification (see below)

04 — agent loop

AI & Agents

Cloud-native platforms are where AI agents actually run. I design the plumbing around the model: retrieval, tools, guardrails, cost and observability, so an assistant can make the jump from demo to production.

Every architecture diagram now has a box labelled "AI". Mine also have arrows going in and out of it.

agent.log

agent run "why is checkout slow?"

  1. plan
    1. 1. read the logs
    2. 2. find the bottleneck
    3. 3. blame DNS
  2. toolmcp.kubernetes.get_pods12 running
  3. toolmcp.datadog.query_latencyp99 2.4s
  4. toolrag.search("checkout runbook")3 sources
  5. toolmcp.coffee.brewpermission denied
  6. answerConnection pool exhausted. Not DNS. (This time.)

  • RAG that cites its sources

    Retrieval-augmented generation on serverless AWS: chunking, embeddings and vector search, with answers grounded in your own documents and linked back to the source. Spend caps and rate limits from day one.

    • Embeddings
    • Vector search
    • Amazon Bedrock
    • DynamoDB

    psst Because even models should read the docs before answering.

  • Agents with a plan and a leash

    Agents that plan, call tools, check their own work and hand over to a human when confidence drops. Designed like any distributed system: timeouts, retries, idempotent tools and a trace of every step.

    • Tool calling
    • Planning
    • Evals
    • Guardrails

    psst Finally, software that procrastinates by planning.

  • MCP: one plug for every tool

    Model Context Protocol servers that give agents typed, least-privilege access to real systems: Kubernetes, cloud APIs, databases and observability. Read-only by default; writes wait for an approval.

    • MCP servers
    • Least privilege
    • Audit log

    psst USB-C for AI tools. Yes, I still carry adapters.

  • Coding agents on the team

    AI agents in the delivery pipeline: code review, migrations, Terraform and Kubernetes changes, incident triage. The agent opens the pull request; a human approves the merge.

    • Claude Code
    • CI/CD
    • IaC
    • Code review

    psst The agent writes the PR. I still get to write "nit:".

05 — credentials

Certifications

13 certifications, collected like Pokémon. Times anyone has asked about one in an interview: 1.

Certifications, newest first
CertificationsIssued
2026
telc Deutsch B1
Language
2025
AWS Certified Security – SpecialtyAWS
AWS
2024
TOGAF® Enterprise Architecture PractitionerTOGAF®
Architecture
Certified SAFe® Product Owner / Product ManagerProduct Owner
Agile
AWS Certified Cloud PractitionerAWS
AWS
2021
Google Cloud Certified – Associate Cloud Engineer
GCP
2020
Certified Calico Operator: Level 1
Networking
Certified Jenkins Engineer
CI/CD
HashiCorp Certified: Terraform AssociateTerraform
IaC
Microsoft Certified: Azure Administrator AssociateMicrosoft Azure
Azure
Microsoft Certified: Azure FundamentalsMicrosoft Azure
Azure
CKA: Certified Kubernetes AdministratorKubernetes
Kubernetes
2018
VMware vSphere 6.5 Foundations
VMware

06 — reach out

Get in Touch

Got an architecture problem? My answer is "it depends", but I'm cheap to hire to explain why.

Have an interesting project or opportunity? Drop me a message and I'll get back to you.

send_message --to manmohan