Tech Lead, AI Agent Platform

Newxel; Warszawa; On-site; PLN 240,000 - 360,000; Full time

Benefits offered by this job

Annual bonus up to 15%

Direct work with CEO

Growth opportunities

Job summary

Newxel in Poland seeks an experienced backend engineer to take over and scale our AI-driven platform. You will own the platform core, runtime, and integration layers, turning the CEO's vision into production-ready systems.

You will lead a growing team, drive secure, cost-conscious releases, and build an Agent Studio in parallel with the core. The role combines hands-on coding with architectural leadership in a fast-moving AI talent company.

Qualifications

• 7+ years backend engineering with Python (asyncio, FastAPI, Pydantic).
• 2+ years leading engineers.
• 1.5+ years building LLM systems in production with real users.
• Strong understanding of state graphs and durable agent execution; LangGraph is a plus.
• Experience with major cloud and infrastructure as code; Google Cloud preferred.
• Ability to take over a system and improve it without rewriting it.

Responsibilities

• Own the platform core: runtime, compiler, memory, APIs and access controls; deliver 30/60/90-day plan with CEO.
• Ensure production platform is fast, reliable and cost-efficient with clear metrics and safe rollbacks.
• Lead Agent Studio development and rollout in parallel with core platform.
• Lead the platform engineering team; plan work, review progress, hire with Recruitment Operations.
• Implement security controls for code, data and AI tools; manage offboarding.

Skills

Python

FastAPI

Leadership

LLM systems

Cloud (GCP)

LangGraph

Tools

Kubernetes

Terraform

GitHub Actions

Job description

Newxel is an AI-native talent solutions company. We provide talent solutions and close engineering roles in the format that works best for the client: from hiring specialists directly into the client's staff, to engaging engineers through our European hubs, to Employer of Record services. Whether it is a single position or a full R&D centre with dedicated labs.

We have been on the market for nine years, with revenue of about $20M, six legal entities, and clients such as Check Point, HiBob and Renesas. Our platform screens 1,000+ candidates in under 10 minutes, and we handle engineer recruitment and hiring ourselves.

Behind this sits our own agent platform: a runtime, a compiler and an organisational memory on Google Cloud. Our business agents run on it in production today: candidate sourcing, CV generation, vacancy creation, policy control and the Copilot our teams work with in Slack. The next step is to turn the platform into a product: client-facing services built on it, and agent integration for other companies.

Why this role exists

The platform works and runs our agents in production. Our CEO designed and largely built it, and today he is the only person who changes its core. At the same time, the platform is becoming a product. We're connecting our first client-facing service to it; a detailed visual Agent Studio is designed and waiting to be built, and the list of improvements is long.

We are hiring an engineer to take over the platform, turn the CEO's architecture and product vision into efficient systems that work in production, and lead the team that builds it.

What you will own

• The platform core. Runtime, compiler, organisational memory, tool integrations, the external API for client-facing products, identity and permissions. In your first two weeks, you study the platform and agree with the CEO on a 30/60/90-day improvement plan. From then on, the plan, its delivery and its outcome are yours.
• Production. A platform that is fast, reliable and economical to run: regular releases without the CEO's involvement, safe rollbacks, clear numbers for failures, stuck runs and cost per run.
• Agent Studio. The visual environment where people design agents, follow their runs, read their memory and manage permissions. It is designed and mocked up on real data. You lead building it and rolling it out, in parallel with the core.
• The engineering team. The platform engineers report to you. You plan their work, review it, decide their access and scope, unblock them and grow them. When a component needs more people, you define the role and hire together with Recruitment Operations.
• Security of code and knowledge. You build and run the controls: access to repositories, documents and data by role; secrets and their rotation; personal data in memory and logs; rules for AI tools on company data; clean offboarding.
• The technical gate for agents. Our agent factory builds business agents as a separate function. You set the platform rules every agent must satisfy and run the technical review before an agent reaches production, regardless of who built it.

How you work with the CEO

The CEO holds the product and architectural vision and moves fast. You take that vision, challenge it with arguments where you see risk, and once a decision is made, you turn it into working code. You do not rewrite the platform to your own taste, and you do not work around it. A short list of architectural principles changes only by written decision from the CEO; everything inside them is yours to decide.

What you will not do

You do not own the product scope or the agent portfolio: those sit with the CEO and the agent factory. You do not write business specifications for agents. You do not change architectural principles on your own.

First 90 days

The outline below is our starting point; you confirm the dates in your 14-day plan. In two weeks, the 30/60/90-day improvement plan is agreed with the CEO. In 30 days, you have made your first production release and written down the release process; our first client-facing service is connected in the sandbox. In 60 days, production runs the same version as the sandbox, releases go out weekly, your team develops and releases the core, the external API is in production, and the access model for code and knowledge is in place. In 90 days, production stability is reported every week in numbers, the first Agent Studio screens are in daily use, and a plan for growing the platform team is agreed.

What we are looking for

• 7+ years of backend engineering with strong Python (asyncio, FastAPI, Pydantic), including 2+ years leading engineers.
• 1.5+ years building LLM systems that run in production with real users and real costs: tool use, streaming, prompt caching, failover between model providers, cost control, quality evaluation.
• You understand state graphs and durable agent execution: what they are and why tools like LangGraph exist. Hands-on LangGraph is preferred, because our whole platform is built on it.
• Strong experience with a major cloud and infrastructure as code. Google Cloud is preferred: all our infrastructure runs on it, and we plan to stay there.
• You work through AI coding agents every day at scale and can show how.
• You have taken over a system built by someone else and made it better without rewriting it.

Please include links to your GitHub or to other work you can show. Share only code and materials you have the right to share.

What we offer

A platform that already works in production with real users, and the mandate to make it a product. Direct work with the CEO who built it, and ownership of the decisions within its principles. A team to lead from day one and the mandate to grow it. A competitive base salary based on your experience, and an annual bonus of up to 15% of base, tied to the platform's goals. Employment in Spain, a B2B or employment contract in Poland, or a B2B contract in Ukraine.