Radu Pirvu

Owned the end-to-end product vision and roadmap for Deloitte's internal GenAI assistant, running on a multi-LLM platform across Azure OpenAI and GCP — with model routing based on use case complexity and data residency requirements.

Defined and delivered a domain-specific RAG agent architecture — purpose-built agents across regulatory compliance, document generation, and ESG frameworks, each backed by a dedicated departmental knowledge base including the firm's global institutional knowledge.

Built and governed a Prompt Library of 15+ categorized, versioned prompt templates — standardizing interactions, reducing time-to-first-useful-output, and enabling consistent quality across use cases. Led delivery of a dedicated agentic workflow tool for presentation drafting, bid/RFP response, and training material generation, with template libraries and visual OCR capabilities.

Implemented a Responsible AI framework — compliance consent, per-session client data classification controls, and data purpose-limitation enforcement — achieving zero regulatory escalations in production.

Tracked real-time token consumption across GCP and Azure; partnered with Finance on capacity forecasting and budget planning. Monitored DAU/MAU adoption metrics and per-prompt satisfaction ratings, translating data signals into sprint priorities that improved productivity outcomes for tens of thousands of practitioners.

Engineered a RAG-based fact-checking solution using Google NotebookLM and prompt engineering to automate source verification.

North Star: 70% processing time reduction, shifting focus toward high-value investigations.

Award: "Most Grounded In Constraints".

Owned the end-to-end lifecycle of the Enterprise CDP, building a unified single customer view (SCV) to enable personalized marketing at scale.

Defined ingestion strategies (streaming/batch), designing ETL pipelines to unify data sources for immediate activation via Braze (CRM). Implemented data quality alerts to minimize errors and ensure reliable data for downstream analytics.

Collaborated with Privacy to establish data quality standards and governance, ensuring full security and strict GDPR compliance.

Partnered with Data Science/CRM teams to deploy propensity models, automating segmentation and predictive targeting for high-value use cases. Managed the Product Backlog and Agile processes, ensuring accurate campaign measurement and clean data flows to Power BI. Promoted data literacy through training, playbooks, and knowledge-sharing documentation.

Scaled the platform to ingest 50M+ daily data events

Drove a 20% revenue uplift by automating AI-driven personalization across email send-time, content, and cadence