# Philipp Noller - AI Consultant & Software Engineer > Freelance AI consultant and engineer: LLM agents, RAG chatbots, GenAI, automation and custom software for startups, SMEs and enterprises - remote across Europe and North America. I'm a freelance AI consultant and software engineer based in Prague, working remotely with startups, SMEs and enterprises across Europe and North America. I design and ship LLM-powered products end-to-end - agentic systems, RAG architectures and multimodal GenAI - backed by solid full-stack engineering and a strong testing and automation background. I care about correctness, reliability and user experience as much as the final result. ## Services - [AI Consulting + Engineering](https://philipp-noller.de/#service): AI pipelines, automations and agents that run in your day-to-day business - not an MVP that looks good in the demo and falls apart a week later. - [Custom Software](https://philipp-noller.de/#service): AI-native CRM, ERP and HR systems built from scratch to fit how you actually work. The software is yours - no licence, no lock-in, no upsell. - [Test Automation & QA](https://philipp-noller.de/#service): ISTQB-certified quality engineering - automated testing for backend, frontend and APIs, performance and load testing, and CI/CD pipelines that keep releases safe. ## Who I work with - Startups: A working AI product in weeks, not quarters: fast iterations, a pragmatic stack and code you own from day one. - SMEs & mid-sized companies: AI for businesses without a data-science department: scoped pilots that pay back, tied into the systems you already run. - Enterprises: At home with procurement, NDAs, security reviews and on-prem or EU-hosted deployment - I have built and led AI platforms inside corporations. - Regions: Remote, in English and German - across Europe and the DACH region as well as the US and Canada. ## FAQ ### What does an AI consultant do - and what do you do differently? Classic AI consulting stops at a slide deck: use cases, a roadmap, a vendor shortlist. I do that part too - which model, which retrieval strategy, build or buy, how to evaluate and operate it - but then I build the system and put it into production. The background makes the difference: I'm an industrial engineer by training and spent 5+ years as a software engineer before AI became my focus, so I know both worlds - the business case and the codebase - and ship production software to a professional standard, not demos. One person from strategy to running code, no hand-off. ### Which companies do you work with? Startups, small and mid-sized businesses and enterprises. The size changes the process, not the standard: a startup gets a working product in weeks, a mid-sized company gets a scoped pilot tied into its existing systems, an enterprise gets someone who has been through procurement, NDAs and security reviews before. ### Where are you based, and do you work with clients outside Europe? I'm a German national based in Prague and work remotely. Most clients are in Germany, Austria, Switzerland and the rest of Europe, and I also work with companies in the US and Canada - in English or German, with overlap hours agreed up front. ### What can you actually build? LLM agents with tool use and MCP, RAG chatbots grounded in your own documents, multimodal GenAI for text, image, audio and video, n8n and custom workflow automation, and the software around it: SaaS platforms, CRM, ERP and HR systems, browser extensions and mobile apps. Production software, not notebooks. ### What else do you offer besides AI consulting and implementation? Custom software for departments and teams - the tools a business would otherwise bend a standard product around: a lean in-house alternative to SAP or Salesforce, an applicant tracking system, a property management or booking system, internal dashboards and back-office tools, built from scratch to fit the actual process. Also Android and iOS apps, browser extensions, web scraping and data extraction projects, and test automation and QA for software you already have. ### Does my custom software have to include AI? No - that is optional. Custom software is a service in its own right: CRM, ERP and HR systems, internal tools and SaaS products are built the same way whether or not a model sits inside them. Where AI earns its place - document intake, search, classification, drafting, assistants - it goes in, and the architecture leaves the door open to add it later without a rebuild. Where it does not, it stays out; nobody should pay for AI as decoration. ### Can our data stay in the EU or on our own servers? Yes. I work with EU-hosted cloud, on-premise deployments and local open-weight models where data must not leave the building, and design for GDPR from the start rather than bolting it on. ### What does AI consulting cost, and how does a project start? Most engagements start with a scoped pilot or proof of concept, so the value is visible before the larger investment. Depending on size and scope - MVP, pilot, production system - typical projects can start at under 5k euros; larger builds with several systems, integrations and ongoing operation sit accordingly above that. The contact form has budget brackets so I can tell you quickly whether the idea fits. ## Selected work ### Marketing AI Platform (2025) One AI platform built solely for marketing: text, image and video generation for Amazon listings, web copy, social posts, ads and blogs, grounded in brand guidelines and company knowledge, with several AI providers and the CMS and DAM under one roof. **The problem** Every new AI capability arrives as its own tool: one for images, one for video, one for copy, another for translation - each with its own login, its own bill and its own idea of the brand. Nothing covers it all, so marketing ends up shuttling assets between systems, and none of them knows the company's guidelines or products. **What it does** - Generates content across text, image and video from a single brief. - Formats the output for the channel it is going to: Amazon listings, website copy, social media posts, ads and blog articles. - Inspirational content - drafts, variations and campaign ideas to start from instead of a blank page. - Grounded in brand guidelines, company rules and product knowledge, so what comes out is on-brand and right for the catalogue. - Several AI providers and APIs under one hood - not only image and video but also text, translation and more - so the best model is used per task and switching providers is configuration, not a rebuild. - Connects to the systems marketing already runs: CMS, DAM and other APIs, so results land where they are used. - Built and rolled out for the business units of an enterprise, from architecture to launch. - Stack: Multimodal GenAI, RAG, LLM APIs, CMS / DAM, Enterprise architecture - Screenshots: https://philipp-noller.de/work/marketing-ai-platform-1.png, https://philipp-noller.de/work/ai-in-marketing.png ### Custom Chatbots (2025 - 2026) Chatbots of every kind - a ready-made platform such as Botpress where that is the right call, and a custom-built secure assistant where it is not, which for most companies is the better deal: your own ChatGPT, on your own infrastructure, without the licence. **Platform or your own build** I build both, so this is a recommendation rather than a sales pitch: for most companies the custom build is the better option. A platform like Botpress stands up quickly, but you pay a monthly plan and usage on top of the model cost, and much of what that buys is product surface a single company bot never touches - while your knowledge base lives on someone else's infrastructure. - A custom assistant is the same chat interface people already know, built for your case only. No licence and no per-seat price; the running cost is the model usage itself. - Secure by construction: it runs on your own servers or EU-hosted, and with open-weight models nothing leaves the building at all - which is what makes internal documents, customer data and regulated material usable in the first place. - It also goes further than a support widget can, because it is your code rather than what a platform chooses to expose: tools, system access and agentic steps. - Ready-made platforms still earn their place - a small public FAQ bot, a team without engineering capacity, a proof of concept due next week. I build those too, and I will tell you which of the two your case actually calls for. **Knowledge** Answers are derived from a RAG knowledge base, databases and internal documents - indexed and retrieved per question, so the bot stays inside your products and your wording. **Capabilities** - Tool use: web search, MCP servers and custom tools. - Image upload, audio and text input. - Connected to the systems around it - hiring, CRM, project boards and brand assets from a single prompt. **Models** Frontier models from OpenAI, Anthropic (Claude) and Google (Gemini), or open-source models such as Llama, Qwen and DeepSeek running locally for full privacy - chosen per use case between data protection, cost and quality. - Stack: RAG, MCP, Botpress, OpenAI / Claude / Gemini, Llama / Qwen / DeepSeek - Screenshots: https://philipp-noller.de/work/custom chatbot - lokal und privat für unternehmensdaten.png, https://philipp-noller.de/work/custom-chatbot.webp, https://philipp-noller.de/work/rag-knowledge-chatbot.webp - Demo video: https://philipp-noller.de/work/custom-chatbot.mp4 ### Cognihire (2024) An AI-native applicant tracking system - jobs, requisitions, candidates and referrals in one workflow, with CVs parsed on upload and a branded careers site per company. **What it does** - Jobs, requisitions, candidates and referrals in one workflow, with a hiring team and stats per posting. - CVs are parsed on upload - the candidate profile is filled in, not typed in. - Multi-tenant: every company gets its own workspace and a branded careers site. - Calendar, reports, e-mail templates and team administration built in. **Browser extension** Puts the ATS on top of a LinkedIn or XING profile: it shows whether the person on screen is already on file and which job they applied for, and syncs the profile into a vacancy or the talent pool in one click without leaving the page. - Stack: ATS, CV Parsing, Multi-tenant, SaaS, Browser Extension - Screenshots: https://philipp-noller.de/work/cognihire.webp, https://philipp-noller.de/work/cognihire-extension-linkedin.webp, https://philipp-noller.de/work/cognihire-extension-xing.webp - Demo video: https://philipp-noller.de/work/cognihire.mp4 ### Custom CRM (2026) An AI-native CRM: leads, accounts and contacts feeding a drag-and-drop opportunity pipeline that keeps probability and forecast up to date as deals move. **What it does** - Leads, accounts and contacts feed an opportunity pipeline; deals are dragged between stages as they progress. - Probability and forecast category update automatically with the stage, and every change lands on the opportunity timeline. - Board and table views, filters per owner and deal, plus inbox, tasks, automations, a knowledge base and reports. - Exposed through MCP, so an AI assistant can create leads, move deals and pull reports from a prompt. **Why custom** No per-seat licence, no roadmap nobody asked for, and a codebase the client owns and can extend - the same argument as the custom software service, shown as a working system. - Stack: CRM, Pipeline, Automations, MCP, SaaS - Screenshots: https://philipp-noller.de/work/custom-crm.webp - Demo video: https://philipp-noller.de/work/custom-crm.mp4 ### myLebenslauf.online (2021 · 2026) A multilingual SaaS résumé platform, designed, built, tested and marketed end-to-end - and rebuilt in 2026 from its legacy code with AI, with MCP compatibility so a résumé can be written straight from ChatGPT. **The service** A résumé and cover-letter builder with free templates: pick a design, fill in experience, education, skills and languages, and download a print-ready PDF. Available in German, English and Spanish. **Built and run solo** Founded and actively operated from 2019 to 2021: full-stack development and testing, UX/UI design, and online and social-media marketing - the whole product, not one layer of it. **Legacy rebuild 2026** The legacy codebase was rebuilt in 2026 with AI tooling on a modern stack, and made MCP-compatible: AI assistants such as ChatGPT can create and edit a résumé through the platform directly from a conversation. - Stack: Full-stack, SaaS, UX/UI, MCP, Legacy rebuild - Screenshots: https://philipp-noller.de/work/mylebenslauf-online.webp - Demo video: https://philipp-noller.de/work/mylebenslauf-online.mp4 ## Experience - 2026 - Present - Self-employed - AI & Software Engineering, Self-employed: AI consulting and implementation - from strategy and architecture through to systems running in production. Alongside that, custom software development: web apps, SaaS products, mobile apps, browser extensions and web design. - 2025 - 2026 - Lead AI Developer, LEDVANCE: Spearheaded the architecture and launch of a marketing AI platform for automated content generation, led multimodal GenAI integration and engineered enterprise-grade RAG chatbots. - 2023 - 2025 - Software / AI Engineer, Stats Perform: Develop and enhance text and vision LLM applications, research agentic AI, MCP and AI coding, and run performance, load and test automation across cloud and on-prem services. - 2022 - Software Engineer - Connected Car, Porsche Engineering: Frontend and backend test automation of hardware-in-the-loop systems with Python and Node.js, CI/CD with GitLab and Jenkins, plus ECU flashing, coding and diagnostics. - 2019 - 2021 - Founder, myLebenslauf.online: Built and ran a multilingual SaaS resume platform - full-stack development and testing, UX/UI design, and online and social-media marketing. - 2019 - 2021 - M.Eng. Smart Factory, Hochschule Esslingen: Master's in smart manufacturing. Thesis: document understanding with AI and RPA, in collaboration with OMM Solutions. - 2014 - 2019 - B.Eng. Int'l Sales Management & Technology, Hochschule Aalen: Engineering degree combining technology with international business, including exchanges in Finland and the USA. ## Tech - AI Coding: Claude Code, Cursor, Copilot - Python: AI, backend & automation - JavaScript / Node.js: Full-stack - LLMs & Agentic AI: Agents, RAG, MCP - Test Automation: Cypress, Playwright, Postman, Selenium - n8n: Workflow automation - AWS / Docker / Kubernetes: Cloud & DevOps - LangChain / LangGraph / CrewAI: Agent frameworks ## Contact - [Contact form](https://philipp-noller.de/#contact): topic (AI Consulting + Engineering / Custom Software / Other) and optional budget - [LinkedIn](https://www.linkedin.com/in/philippnoller) - [Xing](https://www.xing.com/profile/Philipp_Noller5) - Based in Prague, Czech Republic; remote ## Deutsch - [KI-Berater & AI Engineer für den Mittelstand - Philipp Noller](https://philipp-noller.de/de): Freiberuflicher KI-Berater und AI Engineer: LLM-Agenten, RAG-Chatbots, GenAI, Automatisierung und Individualsoftware für Startups, Mittelstand und Konzerne - remote in DACH, Europa und weltweit. ## Pages - [English](https://philipp-noller.de/) - [Deutsch](https://philipp-noller.de/de) - [Sitemap](https://philipp-noller.de/sitemap.xml)