Gig
Hourly, negotiable. Tell us your rate
5
Sep 23, 2026
ABOUT US
We are a small startup (a founder and one developer) with a live web app that helps town planners in New South Wales, Australia. Our stack: Python, Flask, gunicorn and nginx on a Linux VPS, the Anthropic Claude API for report drafting, NSW government spatial services (ArcGIS REST, GeoJSON), and pytest.
THE ROLE
Our developer handles the day-to-day build. We want a senior developer we can call when we hit a hard problem: a short call or a written review to diagnose a bug, check an approach, or advise on design. Our developer then does the implementation. You advise; you don't carry the build.
HOW IT WORKS
- Paid hourly, only for time worked (calls, code review, written advice). No retainer, no minimum.
- The work is irregular. Some months there may be none, some months a few hours.
- We are choosing up to 10 people for an on-call list. If you are on it, we will keep your WhatsApp and
MUST HAVE
- 5+ years of professional Python, including a production web app (Flask, Django or FastAPI) that you maintained over time, not only built.
- Real production experience in at least one of: LLM API pipelines, or GIS/spatial data. Both is a big plus.
- You can explain technical trade-offs in plain English to a non-technical founder, and guide a less experienced developer.
- You can usually reply within 24 to 48 hours.
NICE TO HAVE
Testing and evaluating LLM output, prompt versioning, Shapely/GeoPandas/PostGIS, coordinate reference systems, ArcGIS REST services, Linux server operations.
HOW TO APPLY
Start your subject line with ON-CALL SENIOR. Applications missing any of these items will not be reviewed:
1. Your hourly rate in PHP.
2. Your years of professional Python experience.
3. Your WhatsApp number (with country code) and your
4. Your usual working hours (Philippine time).
5. A link to code you wrote (GitHub, GitLab or similar). If your best work is private or under NDA, say so and describe it instead.
6. Answers to TWO of the three questions below, the two closest to your real experience. 4 to 8 sentences each. Tell us how you would handle it, and where you have dealt with something similar.
A. LLM pipelines: Our code computes planning rules first, then an LLM writes a report that includes a verdict. Sometimes the LLM's verdict contradicts what the code already computed. How would you stop that from reaching users, and how would you test that it stays fixed?
B. Spatial data: We geocode a street address, then look up which zoning polygon it falls in using a government ArcGIS MapServer. Sometimes the point lands on or very close to a polygon boundary. How would you handle that?
C. Flask architecture: A Flask app's main pipeline module has grown very large and is live in production. How would you split it up without breaking the running service?
Please apply by Tuesday 29 September. We review all applications on 30 September.
Thanks,
Kike