A couple of real releases to pin, one data hiccup in the analytics stack, and a set of numbers hiring managers and candidates can actually act on.
Listen to the latest episode and all previous episodes on Snake Signals!
Headline to care about
Polars 1.36.0 was yanked. 1.36.1 is live. If you trialed 1.36.0 anywhere, move to 1.36.1 and re-run any reshapes that touch unpivot or wide-to-long. Conda-forge already serves 1.36.1, so pin versions or you may drift without noticing.
Ship-safe in the last 7 days
Ruff 0.14.9 adds a new RUF100 diagnostic and Pylint tweaks. Update pre-commit so everyone lands the same diffs.
FastAPI 0.124.4 is a clean weekday bump. If you install
fastapi[standard], the Cloud CLI is included, which makesfastapi deployavailable by default.uv 0.9.17 shipped with small quality-of-life fixes. Pin on one service and time CI before and after so you decide with numbers.
Close enough to matter for baselines: Python 3.14.2 and pytest 9.0.2 landed late last week. If you were waiting for stability, add them to your matrix.
The numbers worth using
Pay transparency still moves the needle
Jobs that list pay get a lot more traction. Indeed’s own employer guide cites about 3.8× more applications when a salary is provided. Meanwhile, the UK share of ads with salary fell to 55.3% in October, the lowest since 2021. If your competitors go vague, you get an easy edge by being clear.
What devs actually value at work
Stack Overflow’s 2025 survey puts autonomy and trust, competitive pay, and solving real problems at the top for job satisfaction. Flexibility also matters: 79% of developers prefer hybrid or remote, but only 64% get it. If you want replies, show real ownership, real work, and a clear stance on flexibility.
Developers influence buying decisions
Nearly 48% endorsed or influenced a tooling purchase last year. If your stack is modern and your processes are sane, write that down. People doing the work notice.
AI use is high, trust is not
Adoption keeps rising, but 46% of developers say they distrust AI output. If a take-home uses AI, expect a live review where you explain the code line by line. That’s how loops are catching copy-paste.
One-week plan you can actually run
For teams
Data/ETL: pin Polars 1.36.1 and re-run any melt or unpivot paths. Log before/after shape counts.
API layer: bump FastAPI 0.124.4 mid-week and standardise on
fastapi[standard]if you use the Cloud CLI. Keep the change small and reversible.Tooling: move pre-commit to Ruff 0.14.9. Ship new checks as warnings for one sprint, then enforce.
Build speed: try uv 0.9.17 on a single service. Record “time to green” before and after. Lock only if it pays off.
Metrics to watch this week: median PR CI time, percent of PRs blocked by flakies, lead time from merge to prod. DORA still correlates faster loops with better outcomes, so improve by subtraction first.
For candidates
Top third of your CV: role, stack, impact, scale. Link to code or talks if you can.
Modern Python signals: FastAPI, Pydantic, Ruff, pytest, uv, typed code, async where it helps.
Prove it live: if AI helped, be ready to explain every line. The distrust stat above tells you why.
Use pay data smartly: if an ad lists a band, calibrate your ask inside the range with your scope and impact. If it does not, ask early. The apply-rate gap is on your side.
Closing: what this means
If you are hiring, you win on clarity and speed. Post the band, publish the loop, use a short take-home with a live review, and keep gaps under 48 hours. The market is noisier, but the data says transparency and tight cycles convert.
If you are job hunting, you win on evidence. Show the stack you say you use. Quantify impact. Expect reviewers to probe AI-assisted code. The trust gap is real, so comprehension is currency.
If this helped, pass it to someone who builds or hires in Python. New readers can join at snakesignals.com.
Hiring? Contact
Josh Smith
LinkedIn: https://www.linkedin.com/in/python-recruitment/
Email: [email protected]
Phone: 01727 225 552