Ritesh Puj

Senior software engineer at CrowdStrike. I build the pipelines that bring other companies’ data in, and keep them running.

Berlin, Germany. Building backend systems since September 2013.

Contact me LinkedIn GitHub

Served from a Raspberry Pi in Berlin

  1. Play Games24x7 2013 – 2017, Mumbai Live connections between game tables and players
  2. Gracenote 2017 – 2019, Mumbai An ETL pipeline for a music-rights database
  3. Zalando 2019 – 2022, Berlin Event-driven services for retail operations
  4. CrowdStrike 2022 – now, remote The pipeline that ingests third-party data
FIG. 1Four employers since 2013; the hatched block is the current one. Select a block to read about that role.

What I work on

At CrowdStrike since October 2022, I own the pipeline that ingests third-party data, from architecture to testing. Before that I built backend systems at Zalando, Gracenote and Play Games24x7. I like clean architecture, readable code and tests, fixing broken software and broken processes, and mentoring the people around me.

  1. Third-party sources
    • Azure Event Hubs
    • AWS S3
    • AWS SQS
    • Office 365 Management Activity API
    • Microsoft Graph
  2. Five connectors One per source. Each records checkpoints, so processing resumes where it stopped.
  3. Ingestion pipeline One path into the platform for all third-party data.
FIG. 2The third-party ingestion path at CrowdStrike. I built the hatched block and am responsible for the whole path.

  • Five connectors, one pipeline

    Built the five connectors (204) that bring third-party data in from Azure Event Hubs, AWS S3, AWS SQS, the Office 365 Management Activity API and Microsoft Graph (202). Service owner of two core components the pipeline (206) depends on.

    CrowdStrike, 2022 to now

  • Checkpoints, not duplicates

    Designed and built a checkpoint architecture that makes data processing resumable, eliminated duplicate ingestion and removed the pipeline’s file-size limits. Other teams adopted the approach.

    CrowdStrike, Go and cloud storage

  • Tracing across a department

    Key driver of the OpenTracing and Lightstep integration in my Zalando department, giving its microservices distributed tracing. Also built the task and notification services behind email notifications for Retail Operations.

    Zalando, 2019 to 2022

Experience

The numbers are the blocks in FIG. 1, newest first.

  1. Senior Software Engineer

    CrowdStrike GmbH Remote, Germany

    Promoted to Senior Software Engineer in August 2025; Software Engineer III from October 2022.

    Owns the third-party data-ingestion pipeline end to end.

    Details
    • Responsible for the whole ingestion pipeline for third-party data across the full SDLC, from architecture design to implementation and testing.
    • Built 5 third-party connectors using Azure Event Hubs, AWS S3, AWS SQS, the Office 365 Management Activity API and the Microsoft Graph API.
    • Service owner of 2 core components that support the ingestion pipeline.
    • Designed a checkpoint architecture for resumable processing that eliminated duplicate ingestion; adopted by multiple teams.
    • Co-inventor on US Patent 12,360,962 B1.
    • Observability with Grafana (Graphite as the store), Splunk and LogScale.

    Stack: Go, Python, Apache Kafka, Docker, Kubernetes, Azure Event Hubs, AWS S3, AWS SQS

  2. Senior Software Engineer

    Zalando SE Berlin, Germany

    Promoted to Senior Software Engineer in July 2021; Software Engineer from February 2019.

    Event-driven services for Retail Center, end to end.

    Details
    • Owned the whole development cycle of Retail Center applications, from architecture design to implementation, testing and maintenance.
    • Built REST APIs following Zalando’s open-source API guidelines, and event-driven applications on Nakadi (Zalando’s Kafka wrapper).
    • Built the Task and Notification services that provide email notifications to different domains in Retail Operations.
    • Key driver of the OpenTracing and Lightstep integration in the department.
    • Mentored a junior frontend developer into a full-stack engineer and onboarded several new joiners.

    Stack: Kotlin, Spring Boot, Nakadi, PostgreSQL, Kubernetes, Docker, OpenTracing, Lightstep

  3. Senior Software Engineer

    Gracenote Inc. Mumbai, India

    Designed an ETL pipeline from scratch for a music-rights database.

    Details
    • Designed the architecture of an ETL pipeline from scratch and built it.
    • Developed the core business-logic application for the Music Rights Database project.
    • Led two junior developers through development and implementation.
    • Built a proof of concept Kubernetes cluster with two worker nodes.

    Stack: Java, Spring Framework, ETL, Kubernetes

  4. Software Engineer

    Play Games24x7 Pvt. Ltd. Mumbai, India

    High-traffic WebSocket and TCP services for online games.

    Details
    • Built critical high-traffic tools that keep WebSocket and TCP connections open between game tables and real players.
    • Wrote custom Lua scripts running on Redis to cut an API’s latency.
    • Debugged multithreaded applications with thread dumps and heap dumps.
    • Automated restarts of 60+ applications with shell scripts for deployment.

    Stack: Java, Redis, Lua, WebSocket, TCP, shell scripting, multithreading

United States Patent

(10) Patent No.
US 12,360,962 B1
(45) Date of Patent
Jul. 15, 2025
(54)
Semantic Data Determination Using a Large Language Model
(72) Inventors
Arnd Korn, Erdem Toraman, Nikola Milicic, Ritesh Puj
(73) Assignee
CrowdStrike, Inc.
(21) Appl. No.
18/586,318
(22) Filed
Feb. 23, 2024

Security teams have to analyze data from many third-party sources, and that data rarely arrives with standard field names or complete metadata. This patent describes a system using large language models to automatically determine metadata for fields of non-standardized third-party data.

The LLM describes each field from its name and sample data. A semantic data model framework then refines those descriptions into a finer level of detail, people can validate or override the results, and the descriptions are stored for reuse, so later data can be checked for potential security threats.

My contributions

  • Co-developed the AI system for automated semantic labeling of third-party security data.
  • Designed user override mechanisms for generated field descriptions and semantic models.
  • Implemented comprehensive test coverage for automated threat detection workflows.
Read the patent (PDF) Open on Google Patents

Things I run myself

Both of these live on the same Raspberry Pi that’s serving you this page.

Meal Planner

The app I built to plan meals at home. It’s live, so try it.

  • 200+ vegetarian dishes with ingredients and quantities, searchable by name and category.
  • Cook from what you have: list your ingredients and get dishes ranked by how well they match.
  • Meal calendar to log what you ate, day by day.
  • Jain and vegetarian labels worked out automatically from each dish’s ingredients.

Built with a Go REST API and React with TypeScript, in one Docker container.

Open the Meal Planner

Self-hosted infrastructure

One Go server (306) answers every request to this domain: the site, the meal planner and their APIs. It runs on a hardened Raspberry Pi 4 (300) at home with key-only SSH, fail2ban, rate limiting and automatic updates. A hobby, run like production.

  1. CloudflareDNS and edge proxy

  2. Raspberry Pi 4, Berlin

    1. nginx Reverse proxy
    2. Go container :3000 Site, meal planner, API
FIG. 3The path your request for this page just took.
Ask me about the setup

Skills and background

Languages
Go, Python, Kotlin, Java, Lua, shell scripting
Cloud and infrastructure
Docker, Kubernetes, AWS S3 and SQS, Azure Event Hubs, Microsoft Graph API, Office 365 Management Activity API
Data and services
Apache Kafka, Nakadi, PostgreSQL, Redis, Graphite, ETL pipelines, REST APIs, microservices, Spring Boot, WebSocket, TCP
Observability
Grafana, Splunk, LogScale, OpenTracing, Lightstep
Practice
Clean architecture, testing, the full SDLC, mentoring and team leadership

Education

Bachelor of Science, Computer Science and Engineering
Shah and Anchor Kutchhi Engineering College, Mumbai University. Aug 2010 – May 2013, Mumbai. Score: 64.57%.

Languages

English
Full professional
German
Elementary
Hindi
Native
Gujarati
Native
Marathi
Limited working

Awards

Silver Award, Gracenote (2018)
From the Director of Music Engineering, for critical contributions to the Music Rights project that improved performance and operational efficiency.
Instant Karma Award, Play Games24x7 (2016)
From the Marketing Manager, for automating deployment with a CI tool so designers could publish promotion pages in one click.

Get in touch

Hiring for backend or data-platform work, curious about the patent, or setting up your own self-hosted box? Send a message and it lands in my inbox.

You can also find me on LinkedIn and GitHub.

All fields are required.