Welo Data, the multilingual AI data training division of Welocalize, today announced that it has won two Platinum awards at the 2026 Pinnacle Awards. The Pinnacle Awards recognize groundbreaking innovations and companies across the global technology landscape, honoring excellence and measurable impact across industries.

The company was recognized in: 

• AI Risk & Safety for its Network Identity Management and Operations (NIMO) platform 
• Model Optimization and Compression for its Model Assessment Suite (MAS) 

NIMO was awarded Platinum in AI Risk & Safety for redefining how organizations secure the human layer of AI training. 

As AI systems increasingly rely on globally distributed contributors to generate and evaluate training data, new vulnerabilities have emerged, including identity spoofing, account sharing, bot activity, and coordinated fraud campaigns. For model builders, the stakes are direct: fraudulent contributors corrupt training data at the source, producing models that underperform, hallucinate, or break down under real-world conditions. NIMO addresses these risks through enterprise-grade identity verification protocols, adaptive trust scoring, real-time behavioral analytics, device intelligence, and geolocation validation embedded across the entire data lifecycle.

Unlike traditional post-collection validation approaches, NIMO operates as a proactive prevention layer, detecting and blocking fraudulent activity before compromised data can influence model development. The platform processes approximately 1 million real-time events per month and blocks up to 30 percent of flagged transactions before they enter training pipelines. 

By integrating cybersecurity methodologies inspired by KYC and CIP frameworks into AI training ecosystems, NIMO establishes a new benchmark for responsible AI governance and data integrity.

Welo Data’s MAS received Platinum recognition in the Model Optimization and Compression category for transforming how multilingual large language models are evaluated and improved. 

MAS is a purpose-built evaluation and optimization infrastructure designed to measure multilingual causal reasoning across domains and languages. Moving beyond surface-level accuracy benchmarks, MAS uses expert-authored, domain-specific datasets and structured prompt frameworks to test direct and indirect causal inference while minimizing training contamination risks. 

Since its launch in December 2024, MAS has evaluated 47 large language models across 13 foundation model developers and developed multilingual datasets spanning eight global languages. The framework delivers domain-level diagnostics and actionable insights that support fine-tuning and reinforcement learning from human feedback workflows. 

By turning evaluation into an optimization engine, MAS enables organizations to strengthen reasoning reliability, reduce deployment risk in regulated environments, and make evidence-based model training decisions. 

“Winning Platinum recognition in both AI Risk & Safety and Model Optimization underscores the importance of securing and strengthening AI systems across the full development lifecycle,” said Siobhan Hanna, GM at Welo Data. “Trustworthy AI begins with authentic human data, and it depends on rigorous evaluation that measures real reasoning capability. NIMO and MAS were built to provide the infrastructure that enterprises need to scale AI responsibly and confidently.” 

 Together, NIMO and MAS reflect Welo Data’s integrated approach to AI infrastructure: protecting the integrity of human-generated training data while advancing multilingual reasoning evaluation and model optimization.