Freelance SFT Data Specialist - Secure Code Review - Remote

Role Overview BrainBrowser is a specialised organisation building safety and alignment infrastructure for frontier AI systems. They collaborate with leading AI labs and foundational model developers on supervised fine-tuning data, RLHF pipelines, adversarial red teaming, and safety evaluations. This freelance opportunity places you at the core of their SFT data pipeline, contributing directly to making AI systems safer and more reliable. Key Responsibilities - Review and evaluate AI-generated code samples using a structured 6-point rubric covering prompt realism, vulnerability plausibility, security protocol accuracy, security correctness, functional correctness, and completeness - Assess context and completion pairs for each coding task and evaluate samples against defined acceptance criteria - Flag issues with clear written justification and provide corrected code where required - Distinguish accurately between security bugs and functional bugs, as both are evaluated separately under the rubric Required Qualifications - Minimum 3 years of hands-on software engineering experience - Strong proficiency in at least 3 of the following languages: Python, JavaScript or TypeScript, Java, C#, C, C++, Rust, SQL, or Bash - Solid understanding of security engineering concepts including injection flaws, authentication vulnerabilities, insecure deserialization, path traversal, SSRF, race conditions, and related issues - Clear and precise written communication skills in English - Ability to work independently on a fixed-price per approved violation model with an initial calibration batch of 10 samples Why Join Us This is a unique remote freelance engagement that puts your security engineering expertise to meaningful use in the AI safety space. If you have deep knowledge of secure coding standards and want to contribute to responsible AI development at a global scale, this project offers valuable and flexible work.

- 3 or more years of software engineering experience - Proficiency in at least 3 languages from Python, JavaScript, TypeScript, Java, C#, C, C++, Rust, SQL, Bash - Strong knowledge of security engineering including injection, auth flaws, deserialization, path traversal, SSRF, and race conditions - Ability to distinguish security bugs from functional bugs - Clear and precise written English communication - Familiarity with CWE classifications or security protocols

- Remote work flexibility - Fixed-price per approved violation identified - Opportunity to contribute to global AI safety projects - Engagement with leading AI labs and foundational model developers