Kim Patrick M.

🛡️ Platform safety and data quality are non-negotiable — a single mishandled post or flawed dataset can compromise user trust and brand integrity. 👋 I am Kim, a Trust & Safety Specialist, AI Content Moderator, and Data Annotator with hands-on experience evaluating multimedia content and training machine learning models for enterprise projects. I help tech companies bridge the gap between automated AI systems and high-quality user experiences through precise human-in-the-loop (HITL) review, policy enforcement, and complex dataset evaluation. I offer dual specialization across two core operational pillars: 1️⃣ AI DATA ANNOTATION & RLHF EVALUATION • Frame-accurate video QA: Choppy video/playback timing detection (low frame rate, dropped frames, clock jitter) & shot transition temporal marking (fades, dissolves, wipes, 25-frame zoom rule). • Side-by-side A/B model evaluation for AI visual & anatomical defects (hands, teeth, text rendering). • Precision tagging (bounding boxes, polygon segmentation, keypoints, multi-panel layouts). • RLHF ranking and pairwise model evaluation for LLM and multimodal outputs. 2️⃣ TRUST & SAFETY & CONTENT MODERATION • Severe safety violation review (deepfakes, explicit/NSFW, violence, hate speech). • Child safety policy enforcement and abuse detection. • Fraud, scam, fake account, and copyright policy audits. • Edge-case resolution, false positive/negative reduction, and workflow documentation. 🛠️ Tools & Software: Service Cloud | JIRA Cloud | Labelbox | YOLO | Proprietary AI Annotation & Moderation Dashboards 📈 Track Record & Standards: 📉 Dataset Integrity: Consistently reduced false positives and refined AI training sets. 📝 Clear Documentation: Meticulous edge-case logging and timely operational reporting. 🔒 Reliability: Professional resilience, strict NDA compliance, and reliable delivery. 💬 Ready to strengthen your platform's safety or elevate your AI dataset quality? Send me a message and I will respond promptly!