Software Engineer

Description

🚀 Build the data layer AI needs to do science

Application deadline: 16 Nov 2026 23:59 CET

About Claims

AI ate the Internet, but it can't digest science yet. A paper is thousands of words of prose written for humans, but an AI agent doing research needs to know what was claimed, what evidence supports it, and how far to trust it. Claims turns the scientific literature into exactly that: every claim, tied to its evidence, as structured data that models and agents can query.

We're a startup and run as Subnet 111 on Bittensor. Since our mainnet launch on 28 August 2026, an open competition of miners has processed more than 10,000 papers, and every record was checked by our validator before it entered the dataset. The competition keeps raising the bar: miners keep finding ways to beat our own reference pipeline, and every win becomes part of the standard the whole network is scored against.

The team

Claims was founded by two professors of economics and a senior engineer.

• Philipp Koellinger, CEO, is a professor of economics and an experienced entrepreneur: founder of DeSci Labs and the DeSci Foundation, and co-builder of SciWeave with Ogban.
• Christian Roessler, co-founder, is a professor of economics and game theorist. He designs the incentive mechanism that makes the miner competition work.
• Ogban Ugot, engineering, is a senior backend and machine learning engineer. He built SciWeave, the Claims miner, validator and data pipeline, and you will work with him every day.

You would be the fourth person on the team and our second engineer.

Our culture

We are a fully remote team spread across Europe, Africa and the US West Coast. We have a flat hierarchy and move fast towards clearly defined goals. We think deep, don't shy away from difficult problems or conversations, and we ship relentlessly. The code you will write will run on a live network within your first weeks, and you own real things from day one. We are an AI-native company and expect you to be an AI-native Software Engineer we can trust.

About the role

Position: Software Engineer (Python, full-stack)
Location: Remote. You need to be able to join our team meetings in the afternoon, Central European Time.
Type: Full-time contractor. Start in December 2026 or January 2027.

We're looking for a Software Engineer to help maintain and build Claims and its supporting applications, services, and developer tools. Claims turns scientific research into structured claims and supporting evidence through document processing, AI pipelines, and automated evaluation.

This is a hands-on, backend-focused full-stack role. You'll work across Python services, web dashboards, data pipelines, integrations, and deployment tooling. Your work will combine delivering new features with debugging existing systems, improving reliability, and making the codebase easier to maintain.

You should be comfortable owning well-defined tasks from investigation through implementation, testing, and release, working with the team on larger design decisions and unfamiliar areas.

Responsibilities

• Build and maintain Python backend services, APIs, command-line tools, and background processing workflows.
• Develop and improve web dashboards and internal review tools using TypeScript, React, and modern web frameworks.
• Maintain pipelines that ingest research papers, process documents, extract structured information, and evaluate results.
• Integrate external APIs and AI model providers, handling validation, retries, timeouts, rate limits, and usage costs.
• Work with relational databases and object storage, including schema changes, migrations, queries, and data integrity checks.
• Investigate bugs and production issues across services, pipelines, and interfaces; address root causes and add useful regression coverage.
• Help maintain deployment automation, continuous integration, logging, monitoring, and operational documentation.
• Improve existing code through practical refactoring, dependency maintenance, automated tests, and code review.
• Collaborate with engineering and research colleagues to turn requirements and experimental workflows into reliable software.

Requirements

• Typically 1–5 years of professional software engineering experience, or equivalent practical experience delivering and maintaining production software.
• Strong Python skills, including building maintainable services or data processing applications, managing dependencies, and writing automated tests.
• Experience designing, consuming, and troubleshooting HTTP APIs, with a sound understanding of JSON, validation, authentication, and error handling.
• Working knowledge of TypeScript or JavaScript and experience contributing to React applications or a comparable modern frontend framework.
• Practical SQL and relational database experience, including data modelling, migrations, and debugging queries.
• Confidence working in Linux environments and using the command line to investigate logs, configuration, processes, and service failures.
• Experience with Git, pull requests, code review, automated testing, and continuous integration.
• An understanding of secure engineering practices, including access controls, secrets management, and careful handling of private data.
• The ability to navigate an existing codebase, trace issues across components, and make focused, well-tested changes.
• Clear communication, practical problem-solving, and the judgment to work independently while asking for input when requirements or risks are unclear.

Nice to have

• Experience with FastAPI, Pydantic, PostgreSQL, Supabase, Next.js, or similar tools.
• Experience integrating LLM APIs, building AI-assisted workflows, or evaluating model outputs for accuracy, reliability, latency, and cost.
• Familiarity with PDF parsing, OCR, document extraction, scientific datasets, or evidence provenance.
• Experience with distributed jobs, asynchronous processing, reproducible pipelines, or automated benchmarking.
• Familiarity with Docker, Ansible, systemd, GitHub Actions, cloud hosting, or infrastructure automation.
• Experience with Bittensor, decentralized compute networks, or systems for evaluating independently submitted code.
• Exposure to Rust, native Python extensions, or WebAssembly.
• An interest in scientific research, open-source software, and tools that make research easier to inspect and reuse.

You don't need experience with every technology listed. Strong engineering fundamentals and the ability to learn quickly across the stack matter most.

What success looks like

• You can independently deliver scoped features and fixes across the backend and supporting web applications.
• Your changes are understandable, tested, and documented, with appropriate attention to security and compatibility.
• You help make processing pipelines and production services more reliable and easier to troubleshoot.
• You contribute constructively to reviews, communicate progress and blockers, and help the team turn research ideas into maintainable software.

Benefits

• $1500 - $2000 per month, depending on experience.
• Opportunities for rapid professional growth in a high-impact role: a small team, a live network, and a senior engineer to learn from.
• Quarterly team retreats in inspiring locations (e.g., Vilnius, Koh Samui).
• Travel and visa support for business needs.
• Flexible remote working.

Help us build the data layer that lets AI do science properly.

How to apply

Apply online through the link here: https://airtable.com/appZKkV7FT7eYBhKI/shrjJ3jYfxZjsNckR

It takes about 30 minutes: a few practical questions, four short written answers, links to your code, and your CV or personal website. No cover letter needed.

Applications close on Monday 16 November 2026, 23:59 CET.

We read every application after the deadline and reply to everyone.

Important: To show that you have read this post, include the word Validator in your answer to the question "Why Claims" in the application form.

Please mention the word **WINNABLE** and tag RMmEwNjo5OGMwOjM2MDA6OjEwMw== when applying to show you read the job post completely (#RMmEwNjo5OGMwOjM2MDA6OjEwMw==). This is a beta feature to avoid spam applicants. Companies can search these words to find applicants that read this and see they're human.