Moiz Ahmed

Led the Crane-GPT project, an intelligent assistant to reduce crane operator workload and errors. Productionized BERT-based NLU services using Ray Serve and Python; built Retrieval-Augmented Generation (RAG) pipelines for enterprise search with vector stores and real-time grounding. Delivered a GCP-based containerization project with secure CI/CD and container security policies. Piloted Vertex AI pipelines to automate retraining of product recommendation models, achieving a 25% increase in relevancy during A/B testing. Built dashboards to monitor uptime, latency, and reliability via Grafana; integrated ML models into backend APIs and streaming pipelines using Python and gRPC for real-time recommendations and smart routing.

Led development of Crane-GPT, an intelligent assistant to reduce crane operator workload and errors. Supported research and cross-functional teams to deliver customized ML-driven solutions. Helped balance service quality, cost, and delivery speed through intelligent system design. Collaborated with data science teams to productionize BERT-based NLU services using Ray Serve and Python, reducing inference latency by 30% through optimized batching and model caching strategies. Built and integrated Retrieval-Augmented Generation (RAG) pipelines for enterprise search use cases, leveraging Python, vector stores, and large language models for contextual grounding and real-time retrieval. Delivered a GCP-based containerization project for Richemont’s digital stack, including secure CI/CD pipelines and container security policies. Piloted Vertex AI pipelines to automate retraining of product recommendation models, resulting in a 25% increase in relevancy during A/B testing phases. Worked with

Led development of Crane-GPT, an intelligent assistant to reduce crane operator workload and errors. Supported research and cross-functional teams to deliver customized ML-driven solutions. Helped balance service quality, cost, and delivery speed through intelligent system design. Collaborated with data science teams to productionize BERT-based NLU services using Ray Serve and Python, reducing inference latency by 30% through optimized batching and model caching strategies. Built and integrated Retrieval-Augmented Generation (RAG) pipelines for enterprise search use cases, leveraging Python, vector stores, and large language models for contextual grounding and real-time retrieval. Delivered a GCP-based containerization project for Richemont’s digital stack, including secure CI/CD pipelines and container security policies. Piloted Vertex AI pipelines to automate retraining of product recommendation models, resulting in a 25% increase in relevancy during A/B testing phases. Worked with