Owais M.

Hi, I'm Owais, a US-based AI and data engineer. I build the AI agents, MCP servers, and RAG pipelines that sit on top of a warehouse, and I build the warehouse too: ETL pipelines, dbt models, Snowflake and Redshift. Seven years of this across enterprise and startups, on AWS, Azure, and GCP, including SOC 2 and HIPAA environments. You work with me directly. I do the architecture, write the code, and stay on it until it runs without me. Most of the calls I get now are about AI that's already been bought and doesn't work: a chatbot that answers with confidence and gets it wrong, an agent pointed at the CRM that can't tell which of three lead status fields is the real one. The model is smart enough. What it's missing is everything your team knows without ever writing it down, and it takes the first answer that looks right and stops. Somebody has to put that knowledge where the AI can use it and keep it current. Large engineering teams give that job to a principal engineer. Most companies don't have one. That's the job I do. What I build • AI agents and chatbots that answer from your own data (RAG), with access scoped so each customer or team only sees what it should • MCP servers so Claude, ChatGPT, or your own agents can query your database and internal tools safely • Workflow automation in Python and n8n: intake forms into the CRM, documents and call transcripts into structured records, client data syncs, scheduled reports • LLM pipelines that turn thousands of free-text responses, tickets, or transcripts into fixed categories you can report on • The data layer under all of it: pipelines, dbt models, and a warehouse on Snowflake, Redshift, or Postgres, with the definitions written down • Dashboards and reporting people actually open: Power BI, Microsoft Fabric, ThoughtSpot, Metabase • Rescues: an automation, report, or AI feature someone else built that nobody trusts anymore Some of the work behind that At Upwage, an AI recruiting startup, I was the sole analytics engineer for a year and a half, working directly with the CTO. I built the data platform from scratch, dbt models across 13 sources into Redshift, then a custom MCP server so AI agents could query multi-tenant data with the company scope enforced server-side rather than in the prompt, and a Python pipeline that runs thousands of free-text candidate responses through the OpenAI API into a fixed set of themes the product team can query. I defined the KPIs when no logic existed, and those became how the sales team targeted accounts. ThoughtSpot dashboards for their enterprise clients, including Paramount and Gopuff, sit on top. At Climatebase, a climate jobs platform, I inherited hundreds of undocumented raw tables and consolidated them into a clean, documented dbt project, built the HubSpot funnel and revenue attribution logic, and migrated the warehouse from ClickHouse to Snowflake at 40% lower cost. At a behavioral health practice in Portland I replaced hand-typed weekly referrals and intake follow-ups with production Power Automate flows, HIPAA-aware under the client's BAA. At a Philadelphia nonprofit I built the Microsoft Fabric medallion that pulls their case management and property systems into one set of gold tables. Before going independent I was a data engineer at HP on PySpark pipelines into Redshift, and a Databricks support engineer at Microsoft. How I work Week one is understanding your systems, where the data actually lives, and which question nobody can answer today. Then I build the smallest thing that answers it, put it in front of you, and widen from there. You own the code and the documentation. When a feed breaks, you hear it from me first, with what broke and what I'm doing about it. If you want AI that answers like someone who's been on your team for years, send me two things: what systems your data lives in, and the one question nobody can answer today. I'll tell you what I'd build first. I'm US-based, work US hours, and have 30+ hours a week open.