Meet Philipp Schmid

Summary

Philipp Schmid is an AI developer-experience and developer-relations engineer based in Nuremberg, Germany. He is currently a Staff Engineer at Google DeepMind, where he says he is building the first AI DevRel/DevX team to help bring Google DeepMind research and products to developers.

His stated mission is to help developers build with AI “responsibly, ethically, and successfully,” spanning open models and foundation models such as Gemini and Gemma, across cloud and on-device platforms.

Before Google DeepMind, Philipp was a Technical Lead at Hugging Face, where he led strategic cloud and hardware partnerships with providers such as AWS, Google Cloud, Azure, Cloudflare, DigitalOcean, and Dell. His portfolio says he helped grow Hugging Face cloud and hardware revenue from $0 to about $100M over four years and worked on products such as Hugging Face Inference Endpoints and production LLM deployment tooling.

What he works on

Philipp’s site is primarily a technical blog and project portfolio focused on applied AI engineering. The recurring themes are:

• LLMs and generative AI — fine-tuning, inference, evaluation, RLHF/DPO, RAG, embeddings, and deployment.
• AI agents — especially Gemini agents, managed agents, subagents, evaluation, MCP, tool use, and computer-use workflows.
• Developer experience for AI platforms — tutorials, SDK examples, practical guides, and production deployment patterns.
• Cloud AI infrastructure — Amazon SageMaker, AWS Inferentia, Google Cloud, Azure, Kubernetes, Docker, Terraform, and scalable inference.
• Open-source AI — Hugging Face Transformers, Datasets, Optimum, TRL, Text Generation Inference, and related tooling.

His more recent writing is heavily centered on Gemini APIs and agent systems, including posts such as Credentials API for Gemini Managed Agents, Gemini Managed Agents: Developer Guide, and guides around Gemini Interactions API, MCP, Android computer use, and agent evaluation.

Earlier blog content focuses strongly on Hugging Face and production LLM deployment, for example Introducing the Hugging Face LLM Inference Container for Amazon SageMaker, Llama/Mixtral deployment, SageMaker fine-tuning, AWS Inferentia, Stable Diffusion, Document AI, and optimization of model inference.

Projects and notable work

On his Projects page, Philipp highlights open-source and product work including:

• Deep Learning with PyTorch and Hugging Face — tutorials for PyTorch and Hugging Face libraries.
• Clipper — a Node.js CLI tool for clipping web content and converting it.
• EasyLLM — utilities for working with large language models.
• IGEL — an early German instruction-tuned LLM family based on BLOOM.
• SageMaker Hugging Face Inference Toolkit — maintained work for serving Transformers models on Amazon SageMaker.
• Hugging Face ecosystem contributions, including Transformers, Datasets, Optimum, and TRL.
• Technical leadership around Hugging Face Inference Endpoints, Hugging Face Endpoints on Azure, Hugging Face LLM Inference Containers, and the former Hugging Face Infinity project.

The portfolio website

The website, philschmid.de, is a clean personal technical site with four main areas:

• Blog — a long archive of hands-on AI engineering posts dating back to 2020.
• Projects — selected open-source and product work.
• Newsletter — Cloud Attention, a newsletter-style archive covering AI, LLM, cloud, and infrastructure news.
• About Me — biographical background, role, mission, and technology stack.

The site’s writing style is practical and implementation-oriented: most posts are tutorials, developer guides, or engineering explainers with code examples.

Social profiles

Philipp links to the following profiles from his site:

• GitHub
• LinkedIn
• X / Twitter