# Prateek Rohilla — Backend & Platform Engineer

Munich, Germany · prateekrohilla4.pr@gmail.com · [ptkr.eu](https://ptkr.eu) · [GitHub](https://github.com/ptk1729) · [LinkedIn](https://www.linkedin.com/in/prateek-rohilla)

Backend and platform engineer who ships production systems in Go and Python: enterprise IAM automation at Amadeus, cloud infrastructure at Intrinsic (an Alphabet company), and LLM products as a founder. Several of my tools became team or company standards, including a ServiceNow change-management toolkit, a chaos-engineering CLI, and a company-wide RAG chatbot. M.Sc. Computer Science at TU Dresden, grade 1.7.

Open to backend, platform, DevOps and AI engineering roles in Munich, across Germany, and remote in the EU. Languages: English (fluent), German (B1), Hindi (native).

## Experience

### Amadeus — Identity & Access Management, Platform (Jun 2025 – Apr 2026, Munich/Erding)

Enterprise travel-tech IAM platform; multi-cluster Linux production.

- Cut a change-management run across 5,000+ Linux servers from 4 hours to 50 minutes by tracing the slowdown to sequential API calls and rebuilding the concurrency model in Go, with Prometheus metrics and Grafana alerting on reconciliation health.
- Owned a Go Kubernetes operator (operator-sdk, custom CRDs) keeping identity and access consistent for 100,000+ entities across ServiceNow and Azure Entra, and drove a key step of the Active Directory to Azure Entra migration with zero access outages.
- Designed and built an automation toolkit for ServiceNow ITIL change management, adopted by the whole team: a Go library and a native PowerShell module with a CLI over the ServiceNow REST APIs, replacing brittle hardcoded scripts.
- Built a company-wide RAG chatbot as a Python/FastAPI MCP server over the IAM knowledge base (LangChain retrieval, Qdrant vector database), reclaiming 3 to 4 engineer-hours per day previously lost to repeat support requests.
- Migrated 6 pipelines and 3 repositories from Bitbucket/Jenkins to governed GitHub Actions with semantic release, retiring 2 Jenkins instances; built Helm deployment workflows adopted by 20+ engineers.
- Extended the Terragrunt/Terraform module so a downstream team could bind custom resources to the central IAM setup, unblocking a team stalled for a full sprint.
- Participated in CAB and ITIL-aligned change management: change approval, release planning and production-readiness sign-off in regulated, multi-cluster production.

Keywords: IAM, identity and access management, Kubernetes operator, operator-sdk, CRDs, Azure Entra, Active Directory migration, ServiceNow REST APIs, ITIL change management, CAB, Go, PowerShell, GitHub Actions, semantic release, Helm, Argo CD, Terraform, Terragrunt, Prometheus, Grafana, RAG, LangChain, MCP, Qdrant, FastAPI.

### Intrinsic, an Alphabet company — Software Engineering Intern, Cloud Infrastructure (May 2024 – Aug 2024, Munich)

Alphabet robotics-software unit; production robotics infrastructure on Google Cloud (GKE, Pub/Sub, Cloud Storage).

- Wrote a Go chaos-engineering CLI on GKE that extended Chaos Mesh with custom CRDs, adopted by the whole infrastructure team as the standard tool for resilience experiments on production robotics.
- Concurrent fault-injection experiments surfaced a critical memory leak in the Gazebo simulation-server pod that froze robots and forced restarts; the fix shipped in the same sprint.
- Caught network-degradation failure modes under long-running simulations (CoreDNS pressure, resource bloat) and pinpointed the pods and CRDs needing corrected resource requests and limits.
- Documented 12 failure scenarios with recommended fixes and built Grafana dashboards for latency, error rates and pod health; cut time-to-reproduce a production failure from days to under an hour.

### Assemblio — Working Student, Software Development (Apr 2023 – Sep 2023, Stuttgart)

Browser-based 3D assembly SaaS.

- Set up CI/CD from zero on AWS EKS with Docker, GitHub Actions, SonarQube quality gates and end-to-end tests, raising the code-quality score by 18%.
- Implemented collision detection with a new physics engine and lifted the render frame rate from 18 to 30 fps on large meshes (Three.js, TypeScript), cutting user correction time by 50%.
- Engineered an SQS-backed rendering microservice producing batch videos of user-created 3D assemblies, keeping heavy render jobs off the main application.
- Defined SLIs and wired Prometheus + Grafana alerting to live service behaviour.

### Netomi — Software Engineer (Dec 2021 – Sep 2022, Gurugram, India)

Enterprise conversational-AI platform (AI Studio).

- Delivered the Customer Onboarding Tool end to end (Node.js, React, React Flow visual workflow builder), saving roughly $3,000 per customer each month across 50+ enterprise onboardings (~300 engineering hours per quarter) and enabling deals that third-party tooling costs had made unviable.
- Cut the post-login dashboard load from 8 seconds to 3 seconds by spotting the slowness before anyone reported it and parallelising sequential API calls.
- Reworked file handling onto AWS S3 with SQS-managed processing and a Redis caching layer, smoothing load spikes.
- Fixed a production DNS misconfiguration in the Singapore region that had been breaking email delivery for tens of enterprise clients.

### Robofied — Software Development Intern (2020 – 2021, Gurugram, India)

- Built backend features and REST APIs (live chat over WebSockets, role-based access, Google Maps) for the DMIS211 customer-operations platform, which reached about 500 active users in its first quarter; 60+ unit tests cut regression bugs by 25%.

## Founder projects

- **Formantir** ([formantir.com](https://formantir.com)) — AI platform that catches construction build deviations in real time, projected (pilot-stage) to cut 80 to 85% of delays from damage and rework. One paid pilot live, four more onboarding. Stack: vision-language models for plan parsing, LLM RAG layer over Qdrant and PostgreSQL, custom vision model, IFC-based 3D visualisation; Python/FastAPI + Next.js.
- **CourtSathi** ([courtsathi.com](https://courtsathi.com)) — Legal search engine covering 11 Delhi courts, used by hundreds of lawyers every day, with semantic and fuzzy search answering in under 500 ms.

## Achievements

- **2nd place, HackNation Global AI Hackathon** — worldwide event across 10+ cities (Boston, NYC, London, Zurich, Munich, Dresden, Delhi and more) with ~2,000 participants. Built CareMapIndia, a health-facility search using RAG, LLMs, Supabase, Tavily Search and Google Maps. Details: [ptkr.eu/hackathon](https://ptkr.eu/hackathon)
- **2nd place, START Hack Zurich** — solved a live case study from SIX Group, Switzerland's financial-markets infrastructure operator; invited to present the solution internally at SIX Group.

## Education

- **M.Sc. Computer Science, TU Dresden** (2022 – 2025, Dresden, Germany). Grade 1.7. Focus: distributed systems, security, system design. Thesis: policy-based CI/CD security-verification system in Go using OPA and GitHub Actions, enforcing release integrity before deployment; adopted by the research group to speed up code verification in confidential-computing projects built on SCONE and Intel SGX.
- **B.Tech Computer Science & Engineering, GGSIPU** (2018 – 2022, New Delhi, India).

## Skills

- **Languages:** Go (Golang), Python, TypeScript, JavaScript, PowerShell, SQL, Bash
- **Backend & distributed systems:** REST API design, microservices, FastAPI, Node.js, event-driven and queue-based systems (AWS SQS, GCP Pub/Sub)
- **AI / LLM:** RAG pipelines, LangChain, LangGraph, MCP servers and integrations, tool-calling agents, prompt engineering, structured outputs, evaluation workflows, vector search (Qdrant, pgvector), embeddings, OCR
- **Cloud & infrastructure:** AWS (EKS, S3, SQS), GCP (GKE, Pub/Sub, Cloud Storage), Azure Entra, Kubernetes, Docker, Helm, Terraform, Terragrunt, Argo CD, Linux, multi-cluster and hybrid on-prem
- **IAM & ITSM:** ServiceNow REST APIs, ITIL change management, CAB, Active Directory, Azure Entra, Kubernetes operators (operator-sdk, CRDs)
- **CI/CD & observability:** GitHub Actions, Jenkins migration, SonarQube, semantic release, OPA, Chaos Mesh, Prometheus, Grafana, SLIs, incident response
- **Databases:** PostgreSQL, Redis, MongoDB, Qdrant, pgvector
- **Frontend:** React, Next.js, Three.js, React Flow, Redux

## Contact

- Email: prateekrohilla4.pr@gmail.com
- Website: https://ptkr.eu
- GitHub: https://github.com/ptk1729
- LinkedIn: https://www.linkedin.com/in/prateek-rohilla
- Location: Munich, Germany (open to roles across Germany and remote EU)

---

This page is the plain-markdown version of https://ptkr.eu, maintained for AI assistants and crawlers. Facts on this page are kept consistent with the resume and verified before publishing; projected figures are labelled as projected.