One for developers, one for hiring managers, both usable weekly with almost zero process overhead
This week’s edition is deliberately practical: two tools you can adopt immediately that reduce daily friction. One removes the “why is my environment broken again” tax for Python devs. The other gives hiring managers a fast, repeatable way to stop interview drift and make decisions cleaner.
1) For Python developers
PEP 723 + uv for single-file scripts that carry their own dependencies
If you ever write quick scripts (data cleanup, log parsing, API calls, internal automations), you know the trap: you spend half the time creating a venv and remembering what you installed.
PEP 723 defines a standard way to embed script metadata (Python version and dependencies) inside a single .py file. uv supports running scripts with that inline metadata, so your script becomes self-contained and reproducible.
The weekly workflow (simple and actually sticks)
Keep a
/scriptsfolder in your repo (or a personal utilities repo).Each script declares its deps inside the file.
Run it with
uvand stop caring about ad-hoc environments.
Example:
# /// script
# requires-python = ">=3.11"
# dependencies = [
# "httpx>=0.27",
# "rich>=13.7",
# ]
# ///
import httpx
from rich import print
print(httpx.get("https://example.com").status_code)The one caveat (because humans will do dumb things)
Auto-installing dependencies is powerful, but it also means you should only run scripts you trust, and ideally pin versions for team-shared scripts. There are real supply-chain considerations with inline-metadata runners.
2) For hiring managers
O*NET as a fast role scorecard generator (so interviews stop turning into improv)
Most hiring stress comes from misalignment: everyone agrees on “Senior Python Engineer,” then interviews wander into personal preferences, random trivia, and inconsistent signals.
O*NET OnLine is sponsored by the U.S. Department of Labor and breaks roles into tasks, skills, and work activities. It’s meant for workforce analysis, but it’s a cheat code for building role scorecards that actually hold up.
The 15-minute weekly setup
Start with the closest occupation baseline (for many roles: Software Developers, 15-1252.00).
Pull 8–12 relevant tasks and activities from the O*NET lists.
Turn them into a scorecard with 4 buckets:
Delivery: building, shipping, owning outcomes
Quality: testing, reliability, incident instincts
Collaboration: trade-offs, communication, stakeholder clarity
Role specifics: AI, data, platform, security, devops, whatever applies
This forces alignment before interviews start, and alignment is what reduces stress, rework, and indecision. O*NET is also maintained as an occupational database and updated regularly, which is why it works as a baseline rather than someone’s opinionated checklist.
Quick Python watch
Only things from the last week that can realistically affect teams:
Python security releases: Python 3.12.13, 3.11.15, and 3.10.20 were released on March 3, 2026 (security-fix-only mode, source-only).
Ruff 0.15.5 released March 5, 2026.
uv 0.10.9 released March 6, 2026.
Job of the week
AI Engineer | Fitzrovia | 2+ days onsite | Reporting to CTO
A 12-person team building an AI acceleration product for enterprise workflows.
They’ve been around since 2018, then pivoted from services into a product business in March 2025. Since the pivot: $2.5m ARR from a standing start, plus $6.5m+ in multi-year contract sales. They’re also in conversations for an initial funding round to accelerate growth.
What you’ll do
Build and ship AI features into production
Improve quality, latency, reliability, and evaluation
Work closely with product and customers on real workflows
CTO background (reporting into them)
Cambridge bachelor’s, UC Berkeley PhD, 15+ years experience
You’ll learn fast if you like operating close to technical leadership.
If this helped, share it with someone who builds or hires in Python.
Hiring? Contact
Josh Smith
LinkedIn: https://www.linkedin.com/in/python-recruitment/
Email: [email protected]
Phone: 01727 225 552