Pivoting fully into Python and AI: security, orchestration, and the messy middle

The shift

I am moving my career path toward Python and AI in a deliberate way—not as a buzzword, but as the stack where I build, ship, and break things safely. That means writing more services in Python, leaning on modern LLM tooling, and treating security and orchestration as first-class skills, not afterthoughts.

What I am studying now

AI security

Models and agents are software. They need threat modeling like anything else. I have been digging into the OWASP Top 10 for LLM Applications as a practical checklist: prompt injection, insecure output handling, supply chain issues for models and data, and excessive agency when you give tools to an agent.

For a broader governance lens, NIST’s AI Risk Management Framework is a solid frame for thinking about trustworthiness and deployment risk—not bureaucracy for its own sake, but vocabulary you can use with teams and clients.

Orchestration

“Orchestration” here means multi-step workflows: routing, tools, memory, retries, and human handoffs. I spend time with LangGraph concepts (graphs, state, checkpoints) because they map cleanly to real products—not just one-shot chat completions.

If you prefer a higher-level mental model, LangChain’s overview still helps for chains, tools, and RAG patterns, even when you end up on a slimmer stack.

Python as the spine

None of the above sticks without a comfortable home language. Python.org’s tutorial is the boring answer that works; for APIs I reach for FastAPI because typed models and OpenAPI out of the box match how I like to ship.

What I actually do day to day

A lot of it is glue and judgment: wiring embeddings and retrieval, hardening prompts, logging and evals, and knowing when not to automate. Security and orchestration are the difference between a demo and something you can run in production without losing sleep.

Links quick list

• OWASP LLM Top 10 — https://owasp.org/www-project-top-10-for-large-language-model-applications/
• NIST AI RMF — https://www.nist.gov/itl/ai-risk-management-framework
• LangGraph — https://langchain-ai.github.io/langgraph/
• LangChain docs — https://python.langchain.com/docs/introduction/
• Python tutorial — https://docs.python.org/3/tutorial/
• FastAPI — https://fastapi.tiangolo.com/

If any of this resonates and you are on a similar pivot, feel free to reach out through my site—always happy to compare notes.