Ayush Yadav

Developed and implemented an end-to-end machine learning advertising solution on GCP (Vertex AI, BigQuery ML), leveraging real-time user data for dynamic bidding adjustments, which generated a 28% improvement in Return on Ad Spend (ROAS)

GCP native MLOPS
• Spearheaded scalable Vertex AI pipelines, delivering 30% faster model deployment by leveraging Kubeflow,
scikit-learn, and containerized solutions for enhanced adaptability.
• Architected and deployed a Continuous Training and Continuous Deployment (CT/CD) pipeline with A/B testing,
increasing model selection efficiency by 20%.
• Automated data preprocessing and training workflows using VertexAI and BigQuery, reducing model retraining
time by 10%.
• Engineered and customized a time series forecasting pipeline using Facebook’s Prophet, achieving a 10%
improvement in accuracy for predictive analytics.

GEN AI Document Generation
• Developed an AI-powered documentation solution utilizing LangChain and Azure open AI, reducing manual
documentation effort by 40% and streamlining reverse engineering processes.
• Built APIs for embedding and querying document data using Redis vector databases, enhancing search efficiency
and data usability.
• Orchestrated dynamic task chains with prompt engineering and LangChain, enabling advanced summarization and
Q&A capabilities.

Cloud native data quality
• Developed and deployed a Word2Vec model to enhance data quality within a data pipeline by extracting insights
from user queries, enabling the creation and application of effective data rules and transformations.
• Deployed a containerized solution on Kubernetes, ensuring seamless scalability and reducing infrastructure costs
by 30%.Designed and implemented a cloud-native data quality system, ensuring 50% data accuracy and reliability
across pipelines.