What forward deployed means
I work alongside your team, understand a real workflow, connect the tools and carry the implementation through testing and handover. The starting point is a job your team needs done.
What I build
Custom agents, MCP connectors and repeatable workflows for software teams and agencies. They can gather information, use approved tools and prepare work for human review. The inputs, limits and approval points are defined before building.
Read what I build for examples and boundaries.
How I work
Start with one workflow. Agree what a useful result looks like. Build a limited pilot, test it on real examples, then decide what is ready for production.
Human decisions stay explicit. Client identities and private system details stay private. The working method explains the checks and handover.
The tools behind the work
My work uses Python, MCP, agent frameworks, APIs and databases. The dated technology record lists the stack and when it was reviewed.
From predictive AI to working agents
My public work spans predictive AI and MLOps, generative AI APIs, agent systems and MCP connectors. The 2020–2026 journey traces that progression year by year.
The record counted 111 original repositories: 75 public and 36 private, on 26 August 2026. These are activity counts, not client results. See the sources and counting method, or browse GitHub.
Two systems you can inspect
Raza SOR is my published record: what I build, how I work and the sources behind my claims. People can read the site; connected AI applications can answer from it. Explore Raza SOR.
raza-Developer guides a development job from brief and approved plan through tests, evaluation and a delivery report. See the workflow.
Work with me
Commission an agentic AI system, or bring me into your team as a Forward Deployed Engineer. Both start with one workflow and a conversation about where it gets stuck.