Software Engineer, AI Coding Tasks

If you are an experienced software engineer looking for a flexible way to contribute to the future of artificial intelligence while working from home, Askable is currently seeking specialists to help refine AI coding models. This role focuses on the "evidence layer" of AI, requiring engineers to transform their real-world coding knowledge into complex tasks that challenge current machine learning capabilities.

About the role

This is a remote, contract-based opportunity open to developers globally. You will not be building software products in the traditional sense; rather, you are tasked with creating high-quality coding exercises derived from repositories you are already intimately familiar with. The goal is to design problems that remain difficult for state-of-the-art AI models to solve, effectively helping to "stress-test" and train these systems.

The position offers a flat compensation of US$1,000 per successfully reviewed task. With an expected workload of approximately 10 hours per week, this represents a unique side-hustle for South African developers who want to leverage their professional expertise to earn in foreign currency.

Your daily workflow involves curating coding challenges based on the nuances of public repositories you understand deeply. You will need to take a genuine change or feature from a codebase and convert it into a structured assessment that probes the limitations of AI reasoning.

Key responsibilities include:

• Setting up repository environments within the Askable platform.
• Drafting comprehensive project briefs and defining clear solution criteria.
• Developing automated grading checks to objectively evaluate if a solution is correct.
• Performing "stress tests" on your tasks to ensure they are robust and cannot be easily bypassed by flawed logic.
• Executing tasks against leading AI models and iteratively refining the material based on feedback from internal reviewers.

What they're looking for

Askable requires applicants to have a solid foundation in professional software development. Because you are creating benchmarks for AI, they look for developers who are not only technically proficient but also articulate in explaining complex code changes.

Required skills and experience include:

• At least 2 years of professional software engineering experience.
• Proven contributions to public open-source repositories.
• Advanced proficiency in at least one major language, such as Python, TypeScript, Go, Rust, Java, or C++.
• Strong technical literacy in Linux, Docker, Git, and automated testing frameworks.
• Experience or a strong interest in AI coding tools and evaluating model performance.
• A stable, secure, and private computer setup for high-level development work.

For a South African developer, this role is an excellent way to diversify your portfolio and gain exposure to the mechanics of AI benchmarking. While remote work is standard in the South African tech sector, roles that offer payment in US Dollars are highly sought after. Typically, this type of specialized technical assessment work is found in the niche "AI training data" industry. If you are already active in the local open-source community or contributing to global repos, this position allows you to monetize your existing knowledge of complex codebases without needing to abandon your current full-time employment or standard local consultancy hours.

• 2+ years of professional software engineering experience.
• Code merged into a public open-source repository, ideally one you know well.
• Strong in at least one of Python, TypeScript, Go, Rust, Java or C++.
• Solid with Docker, Linux, Git, and automated testing.
• Regular use of an AI coding tool.
• Experience writing tests, rubrics, or evaluation criteria, ideally for AI models.
• Ability to explain changes clearly and concisely.
• Availability for at least 10 hours a week.
• A secure personal or work computer (no public or shared machines).