Middle Software Engineer (AI-native) — Python
Pinobyte · Full-time · Remote (excluding Russia)
Who we are
Pinobyte is a founder-led engineering company that builds custom data-extraction solutions.
In practice: we get data out of the hardest places on the internet — at scale, reliably — and turn it into something a business can actually use. Marketplaces, e-commerce, real estate, social media and reviews.
We are a compact senior team with no layers of management. You talk to the people who make decisions, and your work ships the same week.
What you'll be doing
You'll be working on our core data-extraction platform: a Python service that plans a request, picks an execution strategy, runs it through the right engine and network path, validates what came back, and reports on everything it did.
You'd own real pieces of it end to end: the API and the async pipeline behind it, the workers that run the load, the execution engines, and the observability that makes a failure explainable.
We'll go through the specifics of the roadmap on the technical call.
Our stack
This is what's actually in the repository, not a wish list:
Language: Python 3.12, asyncio, type hints throughout
Web: FastAPI, Pydantic v2, Uvicorn
Async work: Celery, RabbitMQ, Redis (caching, token buckets, distributed leases)
Data: PostgreSQL, SQLAlchemy 2.0 async, asyncpg, Alembic
Network layer: httpx, curl_cffi, lxml
Headless browsers: Playwright, Camoufox, Patchright, BrowserForge
Infrastructure: Docker, Kubernetes, Helm, Nginx, CI pipelines
Observability: OpenTelemetry, Prometheus, Grafana, Loki, Jaeger
Tooling: uv, ruff, mypy, pytest
You don't need to have used every line of this. You do need to be comfortable reading unfamiliar code and picking things up quickly.
What we're looking for
A solid mid-level Python engineer. Not a tech lead, not an architect — someone who writes good code, ships it, and owns what they shipped.
3+ years of production Python, with real async work behind you: HTTP clients, queues, workers, background jobs.
You're comfortable with SQL and a real database — schema design, indexes, and knowing why a query is slow.
You're not afraid of infrastructure. You can read a Dockerfile, get your service running in a container, find the right log line, look at a dashboard and understand what it's telling you.
You use LLMs as a real part of your workflow — this is a hard requirement. Getting into a large unfamiliar codebase fast, generating code and then reviewing it critically, knowing where the model will confidently lie to you. We work this way every day and we'll ask you exactly how you do it.
You're proactive. When something is broken, slow or badly designed, you come with a concrete fix and trade-offs, not a status update.
You're not afraid to debug — code you didn't write, in production, with incomplete information.
Good spoken English. Not perfect, but enough to discuss a design or walk through an incident on a call.
What we offer
full-time, B2B contract.
Fully remote, flexible hours, async-first — reasonable overlap with the team matters more than a fixed schedule.
Paid AI tooling — Claude Code andwhatever else makes you faster.
Genuinely hard technical problems, a short path from idea to production, and no bureaucracy between you and shipping.
Process
Short and to the point: an intro call, then a technical conversation of about an hour — your past systems, our real problems, how you'd approach them. That's it.
How to apply
Send your CV plus a few sentences on:
a system or service you owned end to end — what broke, and what you did about it;
how you use AI tools day-to-day;
the hardest bug you've debugged, and how you found it.
We read every application and reply to all of them.
Get in touch
Questions, applications, or just want to know more before you commit — reach out either way to telegram or email.