[Contractor] Staff Software Engineer (AI-Native, Full-Stack)

Brazil - Remote | Engineering | Contract | Fully remote

Apply

Cheesecake Labs is an AI, data, and product engineering partner for established and scaling organizations navigating legacy systems, fragmented data, AI pressure, or limited engineering capacity.

Our services include AI Implementation, Application Modernization, Data & Analytics, Product Engineering, AI-Accelerated Delivery, and Embedded Engineering Teams. We deliver agentic workflows in 8-12 weeks and modernize applications in 5-10 months rather than 2 years. We also build trusted data foundations and digital products designed to scale. Our agentic software development lifecycle accelerates execution, while experienced engineers embed with client teams to provide technical depth, security, and clear ownership.

Member of Anthropic's Claude Partner Network; technology partners include Databricks, Snowflake, dbt, and AWS.

Across 13+ years, Cheesecake Labs has built 300+ products for clients including Eurasia Group, MoneyGram, the World Bank, Stellar, Wedgewood, PepsiCo, Merck, Sodexo, and Mercado Libre.

 


The story behind this role

Many AI engineering roles are still built around experiments, isolated prototypes, or research. This is not one of them.

At Cheesecake Labs, we build software and AI-powered solutions for real businesses, with real users, real constraints, and real deadlines. Our squads are small and senior, and they ship in weeks and months, not years.

Today, AI agents write most of our code. This role is for the engineer who decides how things get built, directs the agents, verifies the result, and answers for it in front of the client. At Staff level, that also means deciding how the squads around you specify, build, and verify.

We are looking for a Staff Software Engineer: someone who owns the technical outcome of a project or account, and whose impact reaches beyond their own code. You set how work is specified and verified, shape architecture across projects, and evolve the harness that every squad uses. Where the project has a PM, product direction sits with them and you are their technical counterpart. Where it doesn’t, you take the problem from the founder or client yourself.

 


Engagement details

  • Contractor / project-based engagement
  • Paid in USD
  • Remote-first
  • Flexible allocation depending on project availability and fit
  • Opportunities may vary in duration, scope, and weekly commitment depending on client needs

Some opportunities may start immediately.
Others may become active as new projects enter our pipeline.


What this role actually looks like

This is not a “receive tickets and implement features” role. It is not a “write every line yourself” role either. It is also not a role where you only advise from the side: you stay close to real delivery.

On a typical project, you might:

  • Set how the project specifies work: spec templates, how deep each type of change needs to go, and what a good spec looks like for the client’s domain. Turn recurring review comments into checklists the whole squad uses.
  • Run CKL Spec Driven, our spec-driven development flow, and turn a client’s business problem into a technical design and a spec that an AI agent can execute and you can verify.
  • Design how the project verifies output: what automation checks, what needs human eyes, and which changes always get a second human reviewer, such as auth, payments, and data migrations. Set up automated review and quality gates in the first week of a new project.
  • Direct AI agents (Claude Code, Cursor, or similar) to implement against your spec. Writing code by hand is the exception, not the default.
  • Review what the agents produce, using automated review loops plus your own check of architecture and behavior. You answer for 100% of what ships and can explain every decision in it.
  • Make the architecture decisions across a project or several, and write them down so people and agents can follow them. See where the system will hurt in 6 to 12 months.
  • Lead technical assessments, such as taking over an existing project, and give the Engineering Manager the risk list before kickoff.
  • Be the technical reference when a squad hits a harder area, such as blockchain, DevOps, or data and AI.
  • Own the technical outcome of the project or account. Raise risks to the Engineering Manager early, with options attached, for example when the scope won’t fit the budget.
  • Arrive at client conversations with a recommendation and the technical authority to defend it. Handle the pushback, and when the client makes the call, execute it well. Join pre-sales conversations when technical credibility matters, and bring the opportunities you spot to the Engineering Manager.
  • Evolve the shared harness (skills, agents, and review loops) for the whole company: turn patterns you see across projects into skills, publish them, and teach others how to use them.
  • Mentor Senior and other engineers, and pair with someone who is struggling when the Engineering Manager asks.
  • Keep your own work tested and visible: check it in staging before saying “done”, post short written updates without being asked, and raise blockers the day they appear.
  • Ship and iterate, then close the loop by feeding decisions and edge cases back into the project’s context, so the next spec starts smarter.
  • Work within each client’s rules on AI use (see How we use AI here).

Some projects are heavy on architecture. Others are heavy on client conversations. Most mix both. You work with a lot of autonomy, with an Engineering Manager behind you and squads that count on you.

 


What tends to work well here

People who thrive in this role usually:

  • Have built and shipped real software products end-to-end, and can walk us through the decisions they made, including architecture that held up in production and after the team changed.
  • Already use AI agents daily to design, build, debug, and research, and can show what they built.
  • Think in specs: they turn ambiguity into structured, testable definitions, and can set the standard for how others write them.
  • Have a full-stack mindset and experience across more than one stack. You may be strongest in one layer (backend, frontend, mobile, or AI), but you think about the product as a whole and move across layers when the project needs it.
  • Have raised the level of the people around them, through mentoring, review culture, or tools and standards that other engineers adopted.
  • Have been the technical voice in front of a client: they recommend, defend the recommendation, and still execute the client’s decision.
  • Test what they build before saying “done”, and state clearly what was and wasn’t verified.
  • Surface problems early, with options attached, and share what they learn with the team.
  • Think critically: they question the request, see how a change affects the rest of the system, and bring a recommendation, not only the problem.
  • Work fluently in English with clients and stakeholders (calls, discussions, async), advanced level or above.
  • Operate well with high autonomy, low bureaucracy, and fast feedback loops in a 100% remote setup, knowing when a conversation needs to be live and when async works, and communicating well in both.

 


This probably won’t be your thing if

Just to be transparent:

  • You need detailed requirements or constant direction before getting started.
  • You see AI as a side tool, or you prefer writing most of the code by hand.
  • You are comfortable shipping AI output you haven’t verified.
  • You prefer to keep your impact to your own tasks and your own code.
  • You prefer roles focused only on implementation, without client or business conversations.
  • You are uncomfortable talking directly to clients or working through ambiguous problems.
  • You find it hard to execute the client’s decision after you have made your recommendation.
  • You tend to go quiet between kickoff and delivery, or wait to be asked for an update.
  • You expect the PM or anyone else to test what you built.
  • You prefer stable priorities over fast iteration and high ownership.

There’s nothing wrong with any of that. It’s just not how we operate.

 


One last thing

To participate in this selection process, you must:

  • Have a valid business registration in your country (CNPJ, VAT, EIN, or equivalent)
  • Have a bank account eligible to receive USD payments
  • Be able to issue invoices accordingly

These requirements are necessary for contractor engagement and payment processing.

 


Not sure if you check every box?

That’s okay.

We know that many great people — especially from underrepresented groups — hesitate to apply unless they meet 100% of the requirements.

If this opportunity sounds interesting, we’d still love to hear from you.

IMPORTANT TIP -> The communication during the selection process is done by email. We suggest that you keep an eye on your email inboxes and set the permission to receive messages from @cheesecakelabs.hire.trakstar.com so our messages will not be directed to the spam box ;)

 


How we use AI here

At Cheesecake Labs, AI agents do most of the execution: code, infrastructure, data models, research, prototypes, and documentation. People make the decisions: what gets built and in what order, the architecture, team allocation, and anything about people. Every role here specifies, directs AI, verifies the result, and answers for what ships.

On every project, we also follow a few rules:

  • Use only approved tools, through Cheesecake Labs company accounts
  • Follow each client’s rules on AI. Some clients require a specific AI stack, and highly regulated markets (such as banking in Brazil) may require models hosted in-country
  • Never put client data or personal data (names, birth dates, emails) into an LLM without anonymizing it first. The same goes for our own people’s data
  • Comply with LGPD and other applicable data-protection laws

The collected personal data will be used only to manage and control candidate resumes in order to assess the necessary requirements such as the contact information, and also to conduct this selection process.

The collected data will not be shared with others (“third parties”) without your authorization, and their use will be restricted to Cheesecake Labs internal staff, and also to the service providers that process your data for the selection process. At any time, you can contact us and we will remove your data from our data collection system. In order to do that, just simply send an email to careers@cheesecakelabs.com.