# Faizan Khan

> Muhammad Faizan Khan is an AI and data engineer based in Berlin, Germany, working in data and AI since 2017. He designs and builds AI agents, data platforms and the software around them, on AWS, Google Cloud, Databricks and Snowflake. He has worked independently since the end of 2025.

- Website: https://faizankhan.me/
- Email: hello@faizankhan.me
- LinkedIn: https://www.linkedin.com/in/mfkhan1994
- Location: Berlin, Germany
- Book a call: https://cal.com/mfkhan/30min
- Work: independent; project based, ongoing (a set number of days each month) or advisory

## Services

- AI agents and RAG systems: LLM applications that answer from your own data and take real actions, with testing and monitoring (RAG, Pydantic AI, LLMOps, MLflow).
- Data platforms and lakehouses: new builds or migrations from legacy warehouses, with clean models and governance (Databricks, Snowflake, dbt, Iceberg).
- Pipelines and cloud infrastructure: batch and streaming pipelines set up as code (Airflow, Terraform, AWS, GCP).
- Technical leadership: interim data lead, architecture reviews, data strategy and mentoring.

## Experience

- Sapphire Ventures, Senior Data and AI Engineer (freelance), Sep 2024 to now
- Hella Gutmann Solutions, Lead Data Engineer, Jun 2023 to Dec 2025
- BIT Capital, Senior Data Engineer, Mar 2021 to May 2023
- TrueOcean, Senior Data and Analytics Engineer, Aug 2020 to Feb 2021
- Alten, Data Engineering Consultant (automotive clients including AUDI and MAN), Feb 2019 to Aug 2020
- 10Pearls, Software Developer and Data Engineer, May 2017 to Nov 2018

Certifications: AWS Certified Solutions Architect, AWS Certified Data Analytics, MIT Applied Data Science Program. Education: B.S. Computer Science, FAST-NUCES, Karachi.

## Articles

All articles were published on 2026-10-05. The year shown is the year of the project work each article draws on.

- [The harness matters more than the model](https://faizankhan.me/writing/the-harness-matters): Most of the quality in an AI agent comes from the code around the model. How I think about tools, context, memory and evaluation. (AI Engineering. From building AI agents for clients.)
- [Running AI agents on Google Cloud: keep the tools outside the agent](https://faizankhan.me/writing/ai-agents-on-google-cloud): An architecture for company agents that people use from Slack and Airflow, that query the warehouse, and that stay safe to change. (Architecture. From building an internal agent platform on Google Cloud.)
- [Prompt injection: every input is untrusted](https://faizankhan.me/writing/prompt-injection): Once an LLM can read documents and call tools, any text it reads can try to give it orders. Simple defenses that actually help. (AI Engineering. From building agents that read documents and call tools.)
- [RAG in production is mostly a data problem](https://faizankhan.me/writing/rag-is-a-data-problem): Most bad answers from a RAG system start before the model is even called. Where they come from and how to measure them. (AI Engineering. From putting retrieval over company documents into production.)
- [Prompts are code. Treat them that way.](https://faizankhan.me/writing/prompts-are-code): Small wording changes can move accuracy a lot. Version prompts, test them on a fixed set, and let real code do the maths. (AI Engineering. From shipping LLM features to production.)
- [Data mesh in practice: start with ownership, not tools](https://faizankhan.me/writing/data-mesh-ownership): Data mesh is mostly an agreement between teams. The simple version that worked for me across six teams. (Architecture. From moving data ownership to domain teams.)
- [Moving to a lakehouse without stopping the business](https://faizankhan.me/writing/lakehouse-migration): A step by step way to migrate legacy pipelines to Databricks on AWS while reports keep running. (Data Engineering. From migrating legacy pipelines to Databricks on AWS.)
- [A small, solid data warehouse on AWS](https://faizankhan.me/writing/data-warehouse-on-aws): How I set up S3, Airflow and Snowflake so a growing company could trust its numbers, and what I would still do the same way. (Data Engineering. From building the data warehouse for a fintech team.)

## Optional

- [Full text of all articles](https://faizankhan.me/llms-full.txt)
- [RSS feed](https://faizankhan.me/writing/feed.xml)
- [Sitemap](https://faizankhan.me/sitemap.xml)