Назад
5 часов назад

Python Backend Dev (AI)

Формат работы
remote (только Europe)/onsite
Тип работы
fulltime
Грейд
senior
Страна
UK/Europe
Релокация
UK
vacancy_detail.hirify_telegram_tooltipВакансия из Telegram канала -

Мэтч & Сопровод

Покажет вашу совместимость и напишет письмо

Описание вакансии

TL;DR
Python Backend Dev (AI): Building orchestration runtimes for multi-step AI agents, retrieval and RAG pipelines, and fast asynchronous APIs for wealth management advisors with an accent on Python backend systems and production infrastructure. Focus on designing agent runtimes, integrating vector search and data pipelines, and building reliable async services with AWS.

Looking for a Python Backend Dev.

Folks, an exciting opportunity to join us at an early stage! We're building the future of wealth management, and part of it is already in production.

Remote from the EU or on-site in London — we sponsor UK visas and cover relocation.
Our AI agents are only as good as the Python underneath them.

That’s why I’m hiring Backend Engineers at Nevis. 🐍

We build AI agents that do the operational work of wealth management firms end to end. Last week we closed a $70M Series B led by EQT Ventures, with Charles Schwab joining Sequoia, ICONIQ and Ribbit.

What you’d build:
→ Orchestration and runtimes for multi-step agents
→ Retrieval, RAG and data pipelines
→ Fast async APIs that advisors rely on every day

Stack: Python, asyncio, Pydantic, vector search, AWS.

Why now: a small team, ex-Revolut founders (Mark Swan, Philipp Burda and Ivan Chalov), and ideas that reach production in days, not quarters.

📍 London, or remote anywhere in the EU.

Know a great Python engineer? Tag them below 👇

Link to the role is in the first comment.

#Python #BackendEngineering #AI #Hiring #London
Advisors using Nevis get 8 to 10 hours back every week, and a good part of that starts with not having to write up meeting notes. That’s the area I work on as a backend engineer.
Happy to share our $70M Series B today. Great team, and plenty still to build.
A senior engineer role sitting open for 3 months costs more than the recruiter fee.

Much more. Here's the math nobody does:

Average Senior ML Engineer salary in London: £120,000. In Berlin: €110,000. In New York: $180,000.

That's £10,000+ per month in lost output.
3 months vacancy = £30,000+ in value your product didn't ship.

Add the CTO spending 8 hours a week on interviews instead of building.
Add the team picking up extra load and burning out.
Add the sprint that slipped because you needed one more backend engineer.

Suddenly that open role didn't cost you £0. It cost you £50,000+.

And yet most early-stage startups treat hiring as something to figure out when the pain gets bad enough.

By then, the damage is already done.

The startups I work with across Europe and the US that move fastest on hiring share one mindset:

A vacant senior role is not a saved salary.
It's a ticking cost.

Treat it like one.

Anthony
stevla.com

#hiring #AIstartups #techhiring #Europe #recruitment
Big week for open-weights models 👀 Excited to see what Qwen3.8-Max can do for autonomous coding workflows once it drops.
So true! The "passionate about clean code" line is a total recruiter blindspot now. 😂 Highlighting actual high-load metrics and ownership is how you win the inbox.  🚀
Picking one LLM provider in 2025 is like choosing a single cloud vendor in 2015. You will regret it

Most backend teams treat LLM integration as a point solution. OpenAI works, ship it, move on. Six months later rate limits spike or pricing doubles and you're refactoring across every service that ever called the API directly

No single provider wins across cost, latency, and reliability at the same time. OpenAI's rate limit tiers push you into $1000+/month plans or you queue requests and eat the latency. Claude handles long-context better but costs 3x more per token. Grok is cheap and unreliable. These tradeoffs don't cancel each other out, they shift depending on your workload

We ran a provider-agnostic routing layer at NevoA across 10 microservices. One abstraction, four provider implementations behind it. When OpenAI had a 2-hour outage, traffic auto-routed to Claude with zero application code changes. No incident. No pages. No hotfix deploys at 2am.

The counterargument is real. A routing layer means more code, more test coverage, more observability surface. If your backend makes 10 LLM calls a month, that overhead is not justified

The moment you cross production scale that math inverts. At 10k requests per day, outages and pricing changes are not hypotheticals. They're scheduled pain. The abstraction pays for itself in the first incident it prevents.

Use a single provider if you're prototyping. Use a routing layer if you're shipping to users. The breakeven is around 1000 requests per day

#llm #mlengineering #AI #python #golang #backend #fastapi #softwaredevelopment #techleadership
The shift away from traditional prefix-only caching to a multi-tier storage approach is a really smart move. Being able to reuse text across different serving instances without clogging up GPU VRAM could be a huge unlock for production infrastructure.
This will make integrating Google services into workflows so much easier. The fact that these are fully managed and don't require local setup is a huge time saver.
I’ve recently started exploring the job market again — slowly and without rushing.

The last time I went through interviews was about five years ago, so I expected the experience to feel slightly uncomfortable. Instead, I’m genuinely enjoying it.

There seem to be plenty of Senior Python Engineer and Python Tech Lead opportunities. So far, it doesn’t look as if AI has replaced all software engineers. Maybe the opposite is happening: the growth of AI is creating even more demand for Python, backend systems, integrations, and production engineering.

Another pleasant surprise is how respectful the interviews have been. When I don’t have experience with a particular technology, nobody tries to catch me out or make me feel inadequate. We simply discuss what I know, how I approach unfamiliar problems, and whether our expectations match.

Maybe interviews have changed. Or maybe I was just unlucky five years ago.
Either way, I’m enjoying meeting new people, learning about different products, and seeing how other engineering teams work.

How has the interview experience changed for you in recent years?

#Python #BackendEngineering #TechLead #JobSearch

Будьте осторожны

Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →

Текст вакансии взят без изменений

Источник -