Michael A.Private profile: Upwork shows its earnings only to signed-in members, so we check it every few weeks, not daily. MRR here is a rough estimate (Upwork's six-month earnings ÷ months active). Last read Oct 5, 2026.

AI Model Trainer & Safety Evaluator | Scale AI | Cambridge University I've reviewed thousands of AI outputs. I know the difference between a model that sounds right and one that actually is right and I know exactly what to do when it isn't. I work as an AI Operations & Model Safety Specialist at Scale AI (Outlier AI), one of the most recognized names in AI data training. Day to day, that means auditing Chain-of-Thought reasoning, evaluating high-risk outputs against structured rubrics, detecting hallucinations and sycophantic responses, and escalating systemic failure patterns to engineering teams. I document findings with reasoned logic to support rubric improvement ensuring evaluation standards get sharper over time, not just maintained. I don't just score outputs I diagnose what's going wrong and why. My academic foundation is an M.Phil from the University of Cambridge, where I was trained to analyze complex information under pressure, identify what others miss, and communicate findings with precision. That rigour is what I bring to every evaluation task. What I can help you with: LLM Output Evaluation & Rubric-Based Scoring Chain-of-Thought (CoT) Auditing & Reasoning Quality Assessment AI Safety Review & Bias/Hallucination Detection Training Data Creation & Prompt-Response Annotation Benchmarking & Custom Internal Benchmark Design AI Policy Review & Edge Case Escalation Model Accuracy & Performance Who I work best with: AI labs, foundation model teams, and data services companies who need an evaluator that thinks like a senior specialist not someone working through a checklist. If your model needs to be better than it currently is, let's talk.