Skip to main content

Hire Python developers

Python developers who have shipped more than scripts.

Acasa Labs is a small software team in Surat, India. You can hire a Python developer from the people behind five Python systems we built: NoteTube, Pullsy, Acasa Sentinel, SigmaCPM, and a low-latency options trading platform. That covers web APIs, AI and retrieval, video analytics, and market data, with a senior reviewing every pull request.

Tell us about the roleSee what we built

Trial week first. You own the code.

The problem

"Python developer" covers five different jobs.

The person who writes a good Django view is not automatically the person who can keep a trading engine's hot path free of blocking calls, or tune a detection model to run on a single machine. Most hiring pages promise all of it. We would rather tell you which of those jobs we have done, show you the system, and match you with the developer who did it.

The role

The Python work we hire out

  1. 01

    Web backends and APIs

    FastAPI or Django REST Framework, chosen by what the product needs, with Postgres, migrations, queues, and tests.

  2. 02

    AI, retrieval, and LLM features

    Ingestion, embeddings, hybrid search in Qdrant or pgvector, and model calls with citations and cost tracking, run from a backend service.

  3. 03

    Data and analytics pipelines

    Scheduled pulls from third-party APIs, columnar archives in Parquet, and analysis in Polars and DuckDB that runs in seconds rather than overnight.

  4. 04

    Computer vision on your own hardware

    Detection, tracking, OCR, and appearance search with ONNX and PyTorch models, run locally when footage or documents cannot leave the building.

From production

Python we run, across very different systems.

  • Video analytics with no cloud calls

    Acasa Sentinel

    Acasa Sentinel runs YOLOv8 with ByteTrack, plate OCR in ONNX, and DINOv2 appearance search in Python, with pgvector and PostGIS behind a FastAPI service. All inference is local, so no footage or number plate goes to a third-party API.

  • Python that stays off the hot path

    Options trading platform

    On the options trading platform, market-data decode runs on its own thread and hands off to the engine loop safely; nothing blocking sits on that loop. Research runs on a Parquet tick archive with DuckDB and Polars.

  • Document AI that refuses to guess

    SigmaCPM

    SigmaCPM's Python backend runs OCR on scanned tenders, retrieves candidate pages, and keeps an extracted value only when its quote is found on the cited page.

  • Retrieval scoped per user

    NoteTube

    NoteTube parses, chunks, and embeds uploads into Qdrant with per-user, per-session scoping, then streams grounded answers from a FastAPI service.

Who you get

One developer, and a senior behind them.

We are a small team, not a bench of hundreds. You get a developer who worked on the systems above, and a senior engineer who reviews every pull request before it reaches you. You talk to the developer before you sign, and the trial week shows you how they work in your codebase.

Python skills you can expect

  • FastAPI, Django REST Framework
  • SQLAlchemy, Postgres, pgvector
  • Qdrant, embeddings, reranking
  • Polars, DuckDB, Parquet
  • PyTorch, ONNX, OpenCV
  • asyncio, threading, multiprocessing
  • pytest, Docker

How it starts

A trial week, then your rhythm.

  1. Before day one

    A call to work out which kind of Python job this is. We match you with the developer who built the closest system, and you talk to them first.

  2. Trial week

    One real task in your repository: an endpoint, a pipeline step, or a model evaluated on your data. You get a reviewed pull request and a short write-up.

  3. Weeks 2 to 4

    Your backlog, your tooling, and a daily written update. Anything with numbers, such as latency or accuracy, comes with how it was measured.

  4. Every month

    A review of what shipped and whether the role still matches the work, since Python projects often move from build to data to AI.

Risk-free first week: one developer on one defined task. Continue and the week is billed at the normal rate. Stop within the week and you pay nothing.

Engagement models

Three ways to work with a Python developer.

  • Dedicated developer

    Full time on your product, about 160 hours a month

    An ongoing roadmap and a backlog that will not run out

    $1,500–2,000 a month

  • Part-time developer

    Half time, about 80 hours a month, on fixed days

    A product that is live and needs steady work, not a sprint

    $750–1,000 a month

  • Fixed scope

    A defined piece of work, priced before we start

    A feature, a migration, or an MVP with a clear finish line

    From $1,500, priced per scope

Time zones

How your day overlaps with ours.

We set our working day around yours, with at least 4 hours of overlap every working day, including a daily call for US clients in their morning. India is on UTC+5:30 with no daylight saving. Here is what that means from where you are.

  • UAE

    Dubai

    Your 08:30–17:30 is our 10:00–19:00

    IST is 1 h 30 min ahead

    Your whole working day

    Work like a team in the next building. Calls any time in your day.

  • UK

    London

    Your 09:00–13:00 is our 14:30–18:30 (13:30–17:30 in summer), longer on request

    IST is 5 h 30 min ahead (4 h 30 min in British Summer Time)

    At least 4 hours

    Standup at the start of your day, pairing until your lunch, a written update before you log off.

  • Australia

    Sydney

    Your 13:00–17:00 is our 08:30–12:30 (07:30–11:30 in summer)

    IST is 4 h 30 min behind (5 h 30 min in Australian summer)

    At least 4 hours

    We start early so your afternoon is shared. The work lands before your next morning.

  • US East

    New York

    Your 08:00–12:00 is our 18:30–22:30 (17:30–21:30 in summer)

    IST is 10 h 30 min ahead (9 h 30 min in US summer)

    At least 4 hours, with a call every working day

    A call at the start of your day, and a pull request waiting each morning that was built while you slept.

  • US West

    San Francisco

    Your 08:00–12:00 is our 21:30–01:30 (20:30–00:30 in summer)

    IST is 13 h 30 min ahead (12 h 30 min in US summer)

    At least 4 hours, with a call every working day

    We shift late for your morning: a daily call, then the work continues through your afternoon.

Ownership

You own the code.

  • Repositories sit in your GitHub or GitLab organization from the first commit.
  • Cloud accounts and credentials are yours; we work with access you grant and can revoke.
  • If you stop, nothing is held back: the code, the docs, and the deploy steps stay with you.

Honest trade-offs

When you should not hire a Python developer from us

If the job is model research, training large models from scratch, or a data science team's notebooks, you want a research hire, not a product team. If you need someone in your office, we are remote. And if your product is a mostly TypeScript stack with one Python script on the side, a fixed-scope piece of work will cost you less than a monthly developer.

Proof

Read the case studies before you hire.

Questions

Before you hire a Python developer.

Can one Python developer cover our API and our AI features?

Often, yes, because both run in the same backend. When the AI part grows into retrieval over large document sets or evaluation work, a second person with that experience joins for the design.

Do you do data engineering in Python?

Pipelines that feed a product, yes: scheduled syncs, retries, Parquet archives, and Polars or DuckDB for analysis. We are not a data warehouse consultancy.

Can your developer deploy machine learning models we already have?

Yes. We wrap the model in a service, export to ONNX where it helps, measure latency on your hardware, and add the queue and monitoring around it. Acasa Sentinel runs its models this way.

Which Python version and tooling do you use?

Whatever your project uses, if it is still supported. For new work, a current Python 3 release, type hints, Ruff, pytest, and dependencies pinned in a lock file.

How do you pick the right developer for our project?

By the system they have built, not by a skills list. A trading or realtime project goes to the person who worked on that platform; a document AI project to the one who built SigmaCPM's extraction.

Hire Python developers

Tell us which Python job it is.

API, AI, data, or vision, and what exists today. Trial week first. You own the code.

Tell us about the role

No pitch deck. You own the code from day one.

contact@acasalabs.com

Related