Snr Code Reviewer - C++

About OpenTrain

OpenTrain is a central job hub that aggregates data-labeling and AI-training work from multiple companies and platforms so candidates can find projects in one place. Creating an OpenTrain account is free and applying takes only a few minutes.

We connect skilled reviewers like you with short- and long-term contracting opportunities that help train and evaluate the next generation of AI systems.

About AI training and why it matters

AI models learn from examples prepared and checked by humans: annotating, reviewing, and scoring model outputs so systems become accurate, safe, and useful. This role focuses on the code-review side of that process, ensuring training data about C++ model responses is correct and high quality.

Work like this is often remote and flexible, and it gives you a direct impact on how models behave while letting you apply deep technical skills.

The role

We are seeking a Senior C++ Code Reviewer to audit annotator reviews of AI-generated C++ code responses. You will compile and run submitted snippets inside sandboxed containers, validate they meet the prompt, verify behavior across multiple build configurations, and ensure reviews match our quality rubric.

This is a remote contractor, part-time role requiring 20+ hours per week, paid hourly at $25 USD. The position is open worldwide.

• Employment type: Contractor, Part-time
• Time requirement: 20+ hours/week
• Pay: $25 USD per hour
• Location: Remote — worldwide

What you'll do day-to-day

Your primary task is to validate and correct human annotator reviews of AI-generated C++ code so training data is reliable and safe.

• Compile and execute submitted code in sandboxed/containerized environments to confirm functionality and prompt compliance.
• Test submissions under multiple build configurations and compilers (GCC/Clang/MSVC) and report configuration-specific issues.
• Use sanitizers (ASan/UBSan/TSan), Valgrind, and debuggers to detect UB, memory leaks, and concurrency bugs.
• Assess performance characteristics and recommend optimizations or note regressions using profiling tools (perf, VTune).
• Correct mis-ratings, provide concise constructive feedback, and ensure every review adheres to a structured quality rubric.
• Document reproducible bugs and security concerns (buffer overruns, integer overflows, race conditions) and flag critical issues.

Requirements

Candidates must meet the following non-negotiable requirements drawn from the project brief.

• Experience: 7+ years in professional C++ development, QA, or dedicated code-review roles.
• Modern C++ mastery: Expert knowledge of C++17/20/23 features (constexpr, ranges, concepts), STL, templates, move semantics, RAII, and smart-pointer paradigms.
• Memory & concurrency: Proven ability to detect leaks, undefined behavior, race conditions; strong grasp of threads, atomics, std::async/futures, and lock-free patterns.
• Testing & debugging: Advanced use of GoogleTest or Catch2, ASan/UBSan/TSan, Valgrind, gdb/lldb, and coverage tools.
• Performance optimization: Experience with profiling (perf, VTune), cache-friendly design, SIMD/vectorization, and low-level tuning.
• Secure coding: Familiarity with common C/C++ CWE issues, integer overflow and buffer overrun mitigation, and exploit hardening (ASLR/stack canaries).
• Build & toolchain: Proficient with CMake or Bazel, cross-platform builds, Dockerized toolchains, and CI/CD pipelines (e.g., GitHub Actions, GitLab CI).
• Proof-of-work validation: Comfortable compiling and executing code in sandbox or container environments to verify functionality and prompt compliance.
• Structured QA practice: Experienced with rubric-based scoring, checklists, and ticketing/annotation tools (e.g., Jira, Asana).
• Communication: Excellent written English (B2+ CEFR) to deliver concise, constructive feedback and mentoring notes.

Nice to have

The following are not required but will help you stand out for this role.

• Prior experience evaluating LLM responses, RLHF pipelines, or working on AI/ML data-labeling projects.
• Familiarity with cross-compiler differences and porting code across platforms.
• Experience mentoring reviewers or providing formal code-review training.

How the process works

Apply via OpenTrain: set up a free account and submit your application. If selected, you'll complete a short technical validation that demonstrates your ability to compile, run, and evaluate a C++ snippet in a sandbox.

Work is assigned as batches of submissions to review; you will score each item against a provided rubric, write targeted feedback where needed, and submit corrections. Communication and tracking are done via the platform and integrated ticketing tools.

Because this work trains AI systems, accuracy and clear, actionable feedback are essential. You will be paid hourly at the stated rate for the time you spend reviewing and validating submissions.

• Selection includes a practical code-review test in a containerized environment.
• Assignments are delivered in batches; you must follow the provided rubric for scoring and comments.
• Open to applicants worldwide; contractor relationship will follow platform-specific terms.