München, 27.08.2025

Profile Vladislav Vorobev¶

80469 München
☎ +49 1772597575
✉ j@vavpc.de

AI Agent Engineer & Backend / DevOps – Python | LLM Agents | Kubernetes

Interests and Objective¶

AI Agent Engineer with deep backend roots: 15+ years of Python, distributed systems, event-driven architectures, and Kubernetes, now applied to building production LLM agents and to running engineering teams agent-driven end-to-end. I ship customer-facing chat agents, RAG pipelines, internal tool-using agents, and MCP servers (OpenAI, Claude, LangChain, PydanticAI, FAISS / pgvector). My day-to-day work runs through Claude Code, Cursor, and Zed, backed by a strict Concept → Plan → Spec → Code → Feature workflow and LLM / RAG evaluation discipline. Solid engineering practices — TDD, clean code, SOLID — apply to agents and humans alike. Background spans Automotive, E-Commerce, IoT, Enterprise Audit, and Mapping / Geo.

Last projects / employment¶

Project name or position	Period	Company
Google Distributed Cloud Hosted (GDCH) – DevOps & Hardware	08.2025 – present	Google
AI-Driven E-Commerce Platform – Agents, RAG and ERP Integration	12.2024-08.2025	DesignWeltDeko
SAP Compliance Audit Framework	06.2024-12.2024	SAP
Software Developer Automotive – Battery Testing & Cloud Migration	01.2023-05.2024	Mercedes-Benz
Design and Development of a Distributed Self-Audit Framework	05.2019–12.2022	SAP
Backend Developer – Java, Python & AWS	05.2018-12.2021	Skoobe
Big Data Processing for Automotive Safety Systems – Emergency Brake & Lane Assist	01.2018-05.2020	MAN Truck & Bus
IoT Cloud Platform – Cross-Platform Connectors & DevOps	01.2016-12.2018	Mozaiq
ERP & Analytics Platform – Lead Generation and Online Marketing	01.2016-12.2016	FitnessClubs AV
Global E-Commerce Platform – Localization & Warehouse Management	01.2013–12.2015	…
Before 2015 – Leadership Roles and Early Projects	2009-2014	…

For full descriptions, see the Last projects / employment details section below.

Personal Projects¶

pois.world — Open map & virtual-parcel platform¶

2025 – ongoing

Personal AI-driven project: mark any place on Earth with photos, videos, tags, and chat; own ~1M virtual parcels world-wide on a 2.5D / 3D map. https://pois.world
Built solo, agent-driven from day one (Claude Code, Cursor, Zed) — six microservices, seven shared libs, ~71k LOC.
Concept → Plan → Spec → Code → Feature discipline; spec/feature registry referenced from code and PRs.
Backend: FastAPI + SQLAlchemy 2.x async + GeoAlchemy2 + PostGIS (~1M parcels), Alembic, Pydantic v2.
Frontend: SvelteKit + Svelte 5 (runes), TypeScript, Tailwind, shadcn-svelte, MapLibre GL JS, CesiumJS, Turf.js.
Real-time: unified /ws/live WebSocket multiplexes map events and per-POI chat over Redis pub/sub.
Storage: ScyllaDB (3-node) for chat, MinIO for media, imgproxy for transforms, martin for vector tiles.
AI: pois-mcp MCP server makes the platform first-class to AI agents.
Infra: Docker Compose for dev, Kubernetes for prod.

Tech Stack¶

Languages: Python 3.14, TypeScript
Backend: FastAPI, SQLAlchemy 2.x async, GeoAlchemy2, PostGIS, Alembic, Pydantic v2, httpx
Frontend: SvelteKit, Svelte 5, Tailwind, shadcn-svelte, MapLibre GL JS, CesiumJS, Turf.js, dash.js
Data & Messaging: PostgreSQL/PostGIS, ScyllaDB, Redis (pub/sub + Streams), MinIO
AI / Agents: Claude Code, Cursor, Zed, MCP (Model Context Protocol)
Infra: Docker, Docker Compose, Kubernetes, martin, imgproxy, Overpass, Nominatim

Education¶

University of Applied Sciences of Cologne (FH-Köln)¶

2002–2009

Faculty: Informationstechnik in der Nachrichtentechnik

Title: Dipl.-Ing. Elektrotechnik
Diploma thesis topic: Developed a neural network-based system for image region detection and character classification, implementing Hu Moments for translation, rotation, and scale invariance in C programming.

Lomonosov-University 2005–2006

Faculty of Computational Mathematics and Cybernetics (student exchange program)

Seminars 2005-2008

Negotiation skills, Leave and work in global oriented world, International Project Management across borders, English (WSI), And other…

Telefonbau Nagel GmbH, Cologne / Georg-Simon-Ohm-Berufskolleg 1999–2002¶

Double qualification as IT-Systemelektroniker by Telefonbau Nagel GmbH (now Simply Communicate).

Skills¶

AI / Agents¶

LLM Providers: OpenAI (GPT family), Claude (Anthropic), Hugging Face
Agent Frameworks: LangChain, PydanticAI, custom tool-using agents, Model Context Protocol (MCP) servers
RAG & Vector Stores: RAG pipelines, FAISS, pgvector, Weaviate; embeddings, retrieval, re-ranking
Evaluation: golden datasets, hallucination detection, prompt regression, agent behaviour validation, LangChain evaluation tools, PydanticAI testing utilities
Prompt Engineering: system prompt design, tool use / function calling, structured output, guardrails

Agent-Driven Development¶

Tools: Claude Code, Cursor, Zed, Neovim with AI integrations
Workflow: Concept → Plan → Spec → Code → Feature discipline; spec / plan / feature registry referenced from code and PRs
Project rails: CLAUDE.md and AGENTS.md define design principles, commit conventions, and no-go zones for autonomous edits
Practical use: codegen, refactoring, test scaffolding, code review, architecture sanity checks, runbook authoring, IaC and tooling work

Software Development¶

Programming Languages: Expert in Python (large projects), Groovy/JVM; Proficient in Java, JavaScript, TypeScript, PHP, C/C++; Currently exploring Rust
Frameworks & Libraries (Backend & Data): FastAPI, Pydantic, PydanticAI, FastStream, Pyramid, aiohttp/asyncio, PySpark, Vert.x, Django, Flask, Tornado, Zope/Plone (legacy CMS)
Frontend Development: SvelteKit / Svelte 5 (runes), React, AngularJS, Bootstrap, Tailwind, HTML5/CSS3, Responsive Layouts, Liquid (Shopify Templates), MapLibre GL JS, CesiumJS
Architectural Concepts: Microservices, Event-Driven, Client-Server, TDD, DDD, Clean Code, SOLID, Service-Oriented, REST/gRPC/WebSocket APIs, Data Mesh
Databases: PostgreSQL/PostGIS, MySQL/MariaDB, MongoDB, Cassandra, ScyllaDB, ZODB, Neo4j; ORM: SQLAlchemy 2.x async, GeoAlchemy2, Hibernate, GORM
Big Data & ETL: Apache Spark (PySpark), Airflow, AWS Glue, Databricks
Messaging & Streaming: Apache Kafka (event streaming, distributed log, ETL pipelines), RabbitMQ (message broker, task queues), Celery (distributed tasks), Hazelcast (in-memory grid, pub/sub), Redis (pub/sub, Streams, caching)
Protocols & Formats: JSON, YAML, XML, CSV, Excel, Markdown, Protobuf, Parquet, Arrow, GraphQL, WebSocket, REST

Testing & QA¶

Unit, Integration, Functional, End-to-End, and Regression Testing
pytest, unittest, JUnit, Spock, Behave, Robot Framework
Mocking, Test Automation, Continuous Testing
LLM / RAG evaluation, agent behaviour validation, golden datasets, hallucination detection (covered in detail above under AI / Agents)
OpenTelemetry tracing for observability-driven testing (incl. OpenTracing, OpenCensus, Jaeger)

System Administration, Cloud & DevOps¶

Operating Systems: Linux (Debian, Ubuntu, Alpine, Arch, openSUSE), Windows, iOS
Containers & Orchestration: Docker, LXC, Kubernetes (Calico, MetalLB, Ingress-Nginx, Rook-Ceph), cgroups
Virtualization: KVM, QEMU, libvirt
Cloud Providers: AWS (S3, RDS, Glue, Athena, Databricks), GCP, Azure, Hetzner Cloud, eXtollo
CI/CD & IaC: Jenkins, GitHub Actions, GitLab, Concourse, Terraform, Nomad, Packer, Consul, Vault, Gradle, setuptools, pip, Buildout
Version Control: Git, Mercurial (hg), SVN
Monitoring & Metrics: Prometheus, Grafana, Loki, Tempo, OpenTelemetry, Beyla, Zipkin
Networking & Security: Cisco (Switches/Routers), Bonding, BGP, DNS, SSL/TLS, VPNs, IPSec, iptables/nftables, Wireshark, Nebula, OpenSSL, GnuPG, LetsEncrypt
Web & Infrastructure: Nginx, Apache, HAProxy, Varnish, Memcached, Redis, Samba
Mail & Messaging: Exim, Postfix, SpamAssassin, Dovecot, Courier, Bots, Matrix
Authentication & Directory Services: LDAP & OpenLDAP, Active Directory Integration, Single Sign-On (SSO), OAuth2

Domain Knowledge & Advanced Practices¶

E-Commerce & ERP: Development of custom ERP & E-Commerce platforms from scratch; ERP system integration (250+ employees); Integration with Shopify (APIs, Liquid templates), PayPal, Stripe, custom providers; SEO & Microdata (schema.org), Retargeting & Remarketing (AdTech, SSP, DSP)
APIs & SaaS: Google APIs (Analytics, Merchant, Ads, Identity), OpenAI/Claude APIs, Etsy/Amazon/eBay APIs, Meta/X/LinkedIn APIs, Telegram API, Wikimedia API
Collaboration Tools: Jira, Confluence, Bitbucket, GitHub, GitLab, Redmine, Atlassian Stack
Code Quality & Tooling: PEP8, Ruff, Black, Pylint, Mypy, SonarQube, CodeQL, Clang-Tidy, Checkstyle, SpotBugs, LXML (Python XML Parser)
Modeling & Architecture: UML, Domain-Driven Design (DDD), TDD, Event Bus, Message Queues
Automation & Custom Tooling: Workflow automation for ETL, CI/CD, and infrastructure processes; development of custom scripts and utilities
Data Collection: Large-scale web crawling and scraping with Python (Scrapy, BeautifulSoup, async pipelines)

Non-IT Skills¶

Agile Leadership: Experienced in leading international IT teams (up to 15 members, including remote teams) within Agile frameworks (Scrum, SAFe)
Team & Project Management: Proven expertise in project organization, staffing, leadership, time management, and remote team coordination
Stakeholder Communication: Strong focus on customer satisfaction, safety, and responsible decision-making; experienced in aligning technical goals with business needs

Last projects / employment details¶

Google Distributed Cloud Hosted (GDCH) – DevOps & Hardware¶

08/2025 – present, Munich (consulting)

Company: Google – Role: DevOps Engineer – Type: Consulting Project

DevOps for Google Distributed Cloud Hosted (GDCH) — air-gapped, on-premise Kubernetes for sovereign and regulated workloads
Hardware bring-up and lifecycle automation for distributed cluster fleets
Kubernetes cluster operations, upgrades, and observability tooling
Worked agent-driven where applicable (Claude Code, Cursor) to accelerate runbook, IaC, and tooling work

Tech Stack¶

Infrastructure: Kubernetes, Linux, Bare Metal / Hardware, GDCH, GCP
Automation & Ops: IaC, observability stack, lifecycle automation
AI-Assisted Dev: Claude Code, Cursor

AI-Driven E-Commerce Platform – Agents, RAG and ERP Integration¶

12/2024 – 08/2025, Munich (consulting)

Company: DesignWeltDeko – Role: AI Engineer / Full Stack Developer & DevOps Engineer – Type: Consulting Project

Designed and shipped customer-facing and internal AI agents — Product Chat (RAG over catalogue), Scene Generation, Text Generation, Image Classification — on OpenAI, Claude (Anthropic), LangChain, PydanticAI, FAISS
Worked agent-driven end-to-end (Claude Code, Cursor, Zed) for codegen, refactors, test scaffolding, code review
Built LLM and RAG evaluation pipeline — golden datasets, hallucination checks, prompt regression — to keep agents reliable in production
Connected a custom ERP (without an existing connector) to the Shopify API via an event-driven sync; products, prices, AI-classified images flow in real time
Shipped a Progressive Web App (JavaScript with React and Svelte) and Liquid storefront customisations; integrated marketplaces (eBay, Etsy, Amazon)
Designed and operated Kubernetes clusters (Calico, MetalLB, Ingress-Nginx, Rook-Ceph, Kafka, PostgreSQL) on Hetzner Cloud and on-premise to host platform, vector store, and ERP
Stood up observability and ETL with Prometheus, Grafana, Loki, Tempo, OpenTelemetry, Beyla, Kafka, Airflow
Coached the team on LLM agents, prompt engineering, and AI-assisted workflows

Tech Stack¶

AI / Agents: OpenAI, Claude (Anthropic), LangChain, PydanticAI, RAG, FAISS, custom tool-using agents, LLM evals
AI-Assisted Dev: Claude Code, Cursor, Zed, Neovim
Languages & Tools: Python, JavaScript, SQL, Pandas, Arrow, FastAPI, FastStream, Pydantic, PydanticAI
Frontend: React, Svelte, Liquid Templates (Shopify), HTML, CSS
APIs: REST, GraphQL (Shopify, OpenAI, Google, Etsy, Amazon, eBay), Google APIs (Analytics, Merchant, Ads, Identity)
DevOps & Infrastructure: Linux (Debian, Alpine), Docker, Kubernetes (Calico, MetalLB, Ingress-Nginx), Rook-Ceph, Gitea, Hetzner Cloud ↔ On-Premise, Networking, VPN
Monitoring, Metrics, Tracing, CI/CD & ETL: Prometheus, Grafana, Loki, Tempo, OpenTelemetry, Beyla, Kafka, Airflow (DAGs), Gitea Actions
Testing & Automation: pytest, Behave
Protocols & Formats: JSON, YAML, XML, CSV, Excel, HTML, Markdown, Protobuf, Parquet, Arrow, GraphQL, REST, WebSocket, Liquid

SAP Compliance Audit Framework¶

06/2024 – 12/2024, Walldorf

Company: SAP SE – Role: Software Engineer – Type: Consulting Project

Contributed to feature development for compliance and auditing software
Improved code quality through systematic refactoring of the Python codebase
Upgraded core framework and libraries (Python, Pyramid, supporting dependencies)
Collaborated with SAP teams to ensure maintainability and forward compatibility, Data Mesh
Project follow-up and continuation of earlier work (see Design and Development of a Distributed Self-Audit Framework, 05/2019 – 12/2022, Walldorf)

Tech Stack¶

Languages & Frameworks: Python, FastAPI, Pyramid, SQL
Databases: PostgreSQL, SQLite
Cloud & Infrastructure: Linux, Docker, Kubernetes, Google Cloud Platform (GCP)
APIs & Security: REST API, OAuth, Single Sign-On (SSO)
Testing & CI/CD: pytest, GitHub Actions

Software Developer Automotive – Battery Testing & Cloud Migration¶

01/2023 – 05/2024, Stuttgart

Company: Mercedes-Benz AG – Role: Software Developer Automotive – Type: Permanent Position

Battery test data analysis optimized through microservices-based applications
Cloud migration executed for battery testing applications and workloads to eXtollo / MIC - Azure Cloud (Databricks, VMs, Blob Storage, SQL, Spark)
Data processing pipelines improved for automotive test data using Python, Pandas, Apache Arrow, and Delta Lake
REST APIs (FastAPI) developed and integrated for data access and service communication
Test and analysis results visualized via frontend components built with Svelte
Compliance, safety, and performance standards ensured in collaboration with cross-functional teams

Tech Stack¶

Languages & Frameworks: Python, FastAPI, Pydantic, Multiprocessing, Pandas, Arrow, Svelte
APIs: REST APIs, Azure APIs (Blob Storage, Databricks, Identity, SQL)
Databases & Storage: MongoDB, SQL, Parquet, Delta, CSV, Delta Lake, MDF4 (battery test data format)
Cloud & Infrastructure: Azure (Databricks, Blob Storage, VMs, SQL, Spark), Docker, Linux, Windows
DevOps & CI/CD: Azure DevOps

Design and Development of a Distributed Self-Audit Framework¶

05/2019 – 12/2022, Walldorf

Company: SAP SE – Role: Development Consulting Engineer – Type: Consulting Project

Distributed microservices-based system designed and developed for automated corporate audit processes
Lightweight components implemented to run independently on client systems, collecting and evaluating audit-relevant data states
Alerting system built to notify customers of audit issues, with centralized servers generating reports and visualizations for external auditors
Resilience achieved through components able to handle high loads and network disruptions (offline operation without data loss)
Storage and data management improved with MinIO and cloud storage solutions (AWS, GCP), Data Mesh
PostgreSQL database designed and administered, including performance tuning, backup strategies, data archiving, and monitoring

Tech Stack¶

Languages & Frameworks: Python, FastAPI, Pyramid, Pydantic, asyncio, aiohttp, multiprocessing, threading, pytest
APIs & Protocols: RESTful API, OpenAPI, HTTP, TCP, WebSocket, SSL
Databases & Storage: PostgreSQL, SQLite, MinIO, Cloud Storage (AWS, GCP)
Infrastructure & Messaging: Docker, Linux, Kubernetes, Kafka

Backend Developer – Java, Python & AWS¶

05/2018 – 12/2021, Munich

Company: Skoobe GmbH (Munich, member of the Tolino Alliance) – Role: Backend Developer – Type: Consulting Project

Migration of microservices from Python 2.7 to Python 3, and from Java to Python
Development of asynchronous microservices in Java, Groovy and Python
REST API and GraphQL server development
Migration of monolithic applications to microservices, including DevOps support
Migration of finance reporting system from Python2 to Python3 and to AWS Cloud
Documentation created and maintained for new and migrated services

Tech Stack¶

Languages & Frameworks: Python, Java, Groovy, FastAPI, Django, Tornado, asyncio, Pydantic, SQLAlchemy, Hibernate (Java ORM), LXML, JSON
APIs & Protocols: RESTful API, GraphQL, RPC, OpenAPI, Service Discovery
Databases & Storage: MySQL, NoSQL, SQL
Cloud & Infrastructure: AWS, Docker, LXC, Nomad, Consul, Packer, Ubuntu, Alpine, Nginx
Monitoring & Logging: Prometheus, Grafana, Elasticsearch, Logstash
CI/CD & Automation: Jenkins, HashiCorp stack

Big Data Processing for Automotive Safety Systems – Emergency Brake & Lane Assist¶

01/2018 – 05/2020, Munich

Company: MAN Truck & Bus AG – Role: Big Data / Data Engineer – Type: Consulting Project

Developed Spark-based solutions to process CAN-Bus data (BLF, DBC, MDF4.x) from Emergency Brake Assist and Lane Assist systems for graphical analysis
Designed and implemented distributed Spark jobs to parse, serialize, and classify terabytes of raw sensor data
Developed large-scale PySpark pipelines based on engineering requirements for preprocessing, aggregation, and transformation of automotive safety data
Implemented alerting and trigger mechanisms based on Spark-processed real-time and historical driving data
Developed scalable software and cloud-based pipelines on AWS for storage, processing, and analysis of safety-related data

Tech Stack¶

Big Data & Processing: Apache Spark (PySpark), Hadoop (HDFS), AWS Glue Jobs, Athena, Impala, Hive, Apache Parquet
Databases & Storage: AWS S3, Parquet, MDF4, BLF, DBC (CAN-Bus formats)
Infrastructure & Cloud: AWS, Linux, Docker
Testing & Automation: pytest

IoT Cloud Platform – Cross-Platform Connectors & DevOps¶

01/2016 – 12/2018, Munich

Company: Mozaiq Operations GmbH (ABB, Bosch, Cisco joint venture) – Role: IoT Developer & DevOps Engineer – Type: Consulting Project

IoT platform developed to simplify cross-platform integration of consumer devices and services
Cloud connectors designed and implemented for manufacturers including Bosch/Siemens (BSH), Osram, Philips, Netatmo, and Regardia
Asynchronous, microservices-oriented architecture built with Vert.x and Grails, supporting horizontal scalability and high availability
Platform deployed on Kubernetes (GCP) with Docker, CI/CD pipelines in Concourse and Jenkins
Data security requirements implemented to meet enterprise-grade standards
Interoperability ensured across heterogeneous devices and vendor clouds via the MOZAIQ cloud marketplace
Infrastructure as Code practices established for scalable deployments
Development performed in Scrum team with Java, Groovy, and Python, including automated tests with Spock and pytest

Tech Stack¶

Languages & Frameworks: Vert.x, Grails, Groovy, Java, Python, Node.js, React, Gradle, GORM
Databases & Storage: MongoDB Cluster (replication + sharding), Hazelcast
Messaging & APIs: RESTful API, RabbitMQ
Infrastructure & Cloud: Docker, Kubernetes, GCP, Nginx, Docker Compose
CI/CD & Automation: Concourse CI, Jenkins
Testing: Spock, pytest

ERP & Analytics Platform – Lead Generation and Online Marketing¶

01/2016 – 12/2016, Cologne

Company: fitnessclubs – Role: Software Developer & Architect (Fullstack) – Type: Consulting Project

ERP and analytics platform prototyped and developed for lead collection, processing, and statistical reporting
Responsive UI implemented with Bootstrap for customers and company managers
Existing PHP/Apache-based system migrated to a modern Python stack with data migration to a redesigned database model
System architecture and technical design specified, including ORM-based database modeling
Business logic and partner API integrations (REST) developed for multiple external providers
High-availability cluster and hosting solution designed and deployed on Debian/Linux with Docker and Nginx
Coordination of internal and external staff (IT, UX) throughout project phases

Tech Stack¶

Languages & Frameworks: Python (Python, Pyramid), JavaScript, ReactJS, JQuery, SQL
Frontend: HTML, CSS, Bootstrap
Databases & Storage: MySQL, Redis
APIs & Integration: REST APIs, Partner System Integrations
Infrastructure & Cloud: Debian GNU/Linux, Docker, Nginx, Exim

Global E-Commerce Platform – Localization & Warehouse Management¶

01/2013 – 12/2015, Cologne

Company: [Confidential] – Role: Software Developer & Architect (Full Stack) – Type: Consulting Project

Full e-commerce and ERP platform developed from scratch, with focus on SEO, fast response times, and fully responsive HTML5 layout
Localization and internationalization implemented for layout, payments, taxes, languages, and currencies
Modular architecture designed to support new frontends, e.g., for the Asian market
Initial shop themes delivered for DE, RU, SP, and EU (EN)
Back-office system developed to manage products, sales, and customers
Migration of orders and customer data executed from legacy system to new platform
Close collaboration in a small development team to ensure seamless integration and efficient delivery

Tech Stack¶

Languages & Frameworks: Python, Pyramid, JavaScript, AngularJS v1, SQL
Frontend: HTML, CSS, Responsive Layouts
Databases & Storage: ZODB (File Storage), PostgreSQL
Infrastructure & Hosting: Nginx, Debian GNU/Linux, LXC

Before 2015 – Leadership Roles and Early Projects¶

2005 – 2014

Head of IT Department, leading a team of up to 10 people
CTO at eTargeting (Cyprus), managing a team of up to 10 engineers
Software Development Manager at eTargeting (Cyprus)
Senior Software Engineer at Positiv Multimedia (Germany) – Python (Zope, Plone) and PHP
Startup & experimental GIS project (PostGIS, GeoServer, Leaflet; PostgreSQL/PostGIS administration, replication, backups) – https://mymir.org
Automated competitor data monitoring alerts (project)
Development of retargeting and remarketing systems for online advertising and lead generation
Real Estate Exchange Portal (project)
Hosting Solution (project)
Python Component Framework (project)

Tech Stack¶

Languages & Frameworks: Python (Python), Zope, Plone, PHP, Java, C, ASP, JavaScript
Databases & Storage: PostgreSQL, PostGIS, ZODB, MySQL
GIS & Visualization: GeoServer, Leaflet
Infrastructure & Hosting: Debian GNU/Linux, Nginx, Hosting Solutions, Virtualization (LXC)
Networking & Hardware: Cisco (switches, routers), network bonding, BGP, routing & high-availability setups
AdTech & Data Processing: Retargeting / Remarketing systems, competitor monitoring, data pipelines
Other Tools: Replication & Backup Strategies, Automation Scripts

—

I began professional software development in 2002 with C, ASP, PHP, and Java, before focusing on Python and Zope for full-stack development and IT. Earlier projects and roles before 2014 are omitted here for brevity but can be provided on request.