Hire Yashaswi N. - MLOps Engineer

* Led the development of a comprehensive AI governance policy encompassing Model Access, Accountability, Security, Data Access & Retention, and Model Management. * Architected production infrastructure for LLM embedding and RAG pipelines; optimized chunking and retrieval algorithms, resulting in a 40% improvement in query response times. * Deployed an Agentic LLM in production using Gemini and Langgraph, building a chatbot and recommendation engine that drove a 10% increase in daily user interactions. * Designed load testing strategies for ML and LLM models using Locust, identifying bottlenecks and achieving a 12% reduction in resource consumption during peak periods. * Migrated Kubeflow pipelines to Vertex AI, projecting USD $50,000 in annual cost savings. * Designed and deployed ML pipelines on Vertex AI and Kubernetes, reducing model deployment time by an average of 72 hours. * Migrated Airflow v1 to v2 pipelines, modernizing data processing workflows and improving data management capabilities.

Designed and implemented training pipelines for batch and streaming NLP models (NLTK, spaCy); used MLflow to track experiments, manage configurations, and monitor model performance. - Leveraged AWS EKS to scale and deploy TensorFlow models to production; developed containerization and orchestration expertise across on-premise and OCI Kubernetes clusters. - Automated ML model health monitoring, activity tracking, and deployment pipelines using Ansible, ensuring efficient operations at scale. - Utilized Python, Minio, AWS Lambda, and S3 to optimize AI model storage, deployment, and infrastructure management in a scalable, secure environment.

* Led development of custom Dockerfiles to containerize web applications, backend services, and scikit-learn ML models; deployed containers on a managed Kubernetes cluster on GCP. * Designed and implemented CI/CD pipelines using Jenkins for deploying web applications and ML models (stock prediction and recommendation systems) to GCP. * Established Airflow-based data processing pipelines for batch processing of large-scale stock market datasets. * Built Python automation to evaluate data, generate reports, and manage large stock data dumps in PostgreSQL.