Sonali M

I’m an AI Architect with 12+ years of engineering leadership, specializing in production-grade agentic systems and high-performance Knowledge Graph architectures. I turn complex GenAI PoCs into scalable, cost-efficient enterprise solutions with a strong focus on AI observability and agentic orchestration.

At ABInBev, I lead strategic platform innovation by engineering an organizational Knowledge Graph spanning FE, BE, and ML, cutting developer onboarding ramp-up by 50%. I also spearheaded Graph Summarization for MMM—building a RAG-based engine with Self-Correction Loops to ensure 100% numerical accuracy.

I’ve driven measurable performance improvements across teams, including an 85% reduction in execution latency by optimizing ML infrastructure on AKS/Kubernetes and establishing MLOps standards as an organizational benchmark. I also implement cost-efficient LLM strategies, including API call minimization and cost profiling, while keeping deployments safe, ethical, and reliable through automated evaluation.

Previously at Evalueserve, I solely architected the backend and Dynamic Schema Engine for a global Alzheimer’s research platform—enabling researchers to query any database via GraphQL with zero manual configuration, directly securing $10M+ pharmaceutical contracts. I started my career at Wipro Technologies, where I delivered NLP-driven automation that reduced ticket resolution time by 90% and led cloud transformation from Sybase to AWS Redshift.