Senior C++ Code Reviewer (Sandboxed Audits)

About OpenTrain AI

OpenTrain AI is the platform where people start and grow careers teaching AI. We connect experts with annotation and QA work that directly shapes how next-generation models behave. OpenTrain AI is the hiring and contracting organization for every role.

About AI Training and Why It Matters

AI training (data labeling, annotation, and human-feedback work) is the human foundation of modern AI. Contributors prepare, test, and correct examples so models learn safe, correct, and performant behaviors.

This is flexible, remote work that fits alongside other commitments while putting you on the cutting edge of how state-of-the-art systems are built and improved.

The Role — What You'll Be Doing

You will audit annotator reviews of AI-generated C++ code. For each submission you will compile and run the snippet in sandboxed containers, validate that it follows the prompt, confirm correct behavior under multiple build configurations, and verify adherence to performance and security best practices.

When annotator ratings are incorrect or incomplete you will correct scores, add targeted feedback, and ensure every review meets our quality rubric so the labeled data is reliable for model training.

• Compile and execute code safely in containerized sandboxes to confirm prompt compliance.
• Validate across build configurations and compilers (GCC/Clang/MSVC) when applicable.
• Detect undefined behavior, memory leaks, race conditions, and security issues.
• Produce concise, constructive feedback and correct mis-ratings according to the rubric.

Core Responsibilities

This role focuses on hands-on code validation, security and performance checks, and high-quality written feedback. You will follow structured QA practices and use established toolchains and CI patterns to reproduce and verify issues.

• Run sanitizers (ASan/UBSan/TSan), Valgrind, and address sanitizer outputs to find UB and memory issues.
• Use unit-test frameworks (GoogleTest/Catch2) and debugging tools (gdb/lldb) to validate behavior.
• Profile performance with tools like perf or VTune and evaluate cache-friendliness and vectorization where relevant.
• Apply rubric-based scoring, checklists, and ticketing workflows (Jira/Asana) for traceable QA.
• Document fixes, suggest secure coding mitigations (e.g., integer overflow and buffer-safety), and mentor annotators by example.

Requirements — Must Haves

These requirements are drawn exactly from the role description and are essential to perform the work safely and effectively.

• 7+ years in professional C++ development, QA, or dedicated code-review roles.
• Expert knowledge of modern C++ (C++17/20/23): constexpr, ranges, concepts, STL, templates, move semantics, RAII, and smart pointers.
• Proven ability to detect memory leaks, undefined behavior, race conditions; strong understanding of threads, atomics, std::async/futures, and lock-free patterns.
• Advanced experience with GoogleTest or Catch2, sanitizers (ASan/UBSan/TSan), Valgrind, and gdb/lldb.
• Skilled with profiling and low-level performance tuning across compilers (perf, VTune, SIMD/vectorization).
• Familiarity with secure-coding issues for C/C++ (CWE/OWASP), exploit mitigations, and common vulnerability patterns.
• Proficient with build systems and toolchains (CMake/Bazel), Dockerized toolchains, and CI/CD pipelines.
• Comfort compiling and executing code in sandbox/container environments to verify functionality and prompt compliance.
• Structured QA experience: rubric-based scoring, checklists, and ticketing/annotation tools.
• Excellent written English (B2+ CEFR) and ability to deliver concise, constructive feedback.

Nice To Have

These skills are optional but will make you especially effective in this role and helpful to the team.

• Background in LLM evaluation, RLHF pipelines, or prior AI/ML data-labeling projects.
• Experience with cross-platform portability issues and Windows-specific toolchains (MSVC) in large codebases.
• Prior mentoring or formal code-reviewer responsibilities in a high-volume QA environment.

How It Works, Schedule, and Compensation

This is a remote, contract, part-time role with flexible hours. You will be paid per hour and expected to work roughly 20+ hours per week. OpenTrain AI manages contracts and payments for all contributors.

Compensation: PAY_PER_HOUR at USD 25/hour as listed in the role details. The role is open worldwide; you must be able to perform sandboxed builds and testing from your environment and follow our secure remote execution guidelines.

• Time requirement: 20+ hours/week (flexible scheduling).
• Payment: USD 25 per hour (PAY_PER_HOUR).
• Worldwide applicants accepted; ensure reliable internet and a secure development environment for sandboxed testing.