Paul Dave

Senior AI Full Stack Engineer specializing in LLM-powered applications, RAG architectures, and intelligent data systems. 10+ years of experience building scalable platforms using React, Node.js, Python, and AWS, with a strong focus on integrating AI into real-world products. Experienced in designing low-latency, data-driven systems that turn complex data into actionable insights and automation.

Senior AI Full Stack Engineer at Mercury (2023-01 – Present)

• Led development of Mercury Insights, an LLM + RAG-based financial intelligence platform enabling natural language queries over financial data for 200K+ customers, reducing manual reporting effort by 60%.
• Built predictive cashflow intelligence engine, analyzing millions of transactions daily to generate real-time alerts on burn rate and anomalies, improving financial risk detection by 40%.
• Architected AI services (Python, FastAPI, AWS Lambda) integrating vector embeddings and retrieval pipelines, delivering low-latency conversational responses for financial workflows.
• Designed secure AI data-access layer using OAuth and API gateways to enable agentic workflows between LLMs and financial systems with full auditability.
• Developed ML-based transaction categorization system with feedback loops, improving classification accuracy by 30%+ over time.
• Built real-time financial dashboards (React, Next.js, GraphQL) handling millions of records, improving user engagement by 20%.
• Implemented event-driven architecture (Kafka, AWS SNS/SQS) processing 100K+ financial events/hour for alerts and automation workflows.
• Optimized PostgreSQL and AWS RDS queries, scaling throughput to 5M+ daily records while maintaining sub-300ms latency.

Senior Software Engineer at Uber (2018-10 – 2022-12)

• Contributed to Michelangelo ML Platform supporting 10M–30M real-time predictions/sec, powering ETA, fraud detection, and marketplace optimization at global scale.
• Built distributed ML pipelines and feature systems, enabling reuse across 5,000+ production models and accelerating model deployment cycles.
• Developed real-time event streaming systems (Kafka, Redis, Spark) processing 400K+ events/min, improving dispatch and matching efficiency by 6%.
• Designed scalable microservices (Java Spring Boot, Node.js) supporting 40M+ daily ride requests with high availability and low latency.
• Built real-time analytics dashboards (React + backend services) achieving 10K RPS throughput and reducing latency by 30%.
• Integrated ML inference into production systems for fraud detection and operational decisioning, improving system responsiveness and reliability.
• Improved observability pipelines (Datadog, logging, metrics), reducing incident resolution time by 40%.

Frontend Engineer at Chewy (2015-01 – 2018-08)

• Built and enhanced customer-facing eCommerce experiences using React, TypeScript, JavaScript, HTML, and CSS, improving browsing, product discovery, and checkout usability across web platforms.
• Developed reusable UI components and frontend architecture patterns, helping teams ship features faster while keeping the user experience consistent across Chewy's online storefront.
• Integrated frontend applications with REST APIs and backend services, enabling real-time product data, order updates, and personalized shopping experiences for millions of customers.
• Optimized page performance, responsiveness, and accessibility across key shopping flows, reducing load times and improving engagement on high-traffic pet retail pages.

Bachelor of Science (B.S.) in Computer Science – The University of Texas at Dallas (2011-09 – 2014-05)