Senior Director of Generative AI and Head of Developer Experience at Bloomberg LP, leading AI infrastructure deployment across 9,000+ engineers. Brings practical insights from implementing AI tools and agent systems in a mission-critical financial services environment with extensive brownfield systems.
Champion of Strategic AI Deployment & Developer-Centric Tools
Lei Zhang advocates for thoughtful, strategic AI deployment in enterprise environments, challenging the industry’s tendency toward blanket AI adoption. His philosophy: match tools to problem types, eliminate undesired work, and treat AI deployment as organizational change management.
Current Work
As Senior Director at Bloomberg (since November 2017), Lei leads the Developer Experience group within Software Infrastructure Engineering and oversees generative AI initiatives. His team focuses on:
- Strategic AI deployment - Identifying where AI creates value vs. accelerates technical debt
- Agent systems - Building uplift agents for code improvements and incident response agents for alert management
- Organizational change - Running 20+ training programs treating AI adoption as change management
- Enterprise infrastructure - Supporting 9,000+ engineers, 2,000 data specialists, 400+ ML/AI professionals
Previously spent 10 years at Bloomberg progressing from software engineer in BVAL through Derivatives group leadership roles.
Key Insights from Bloomberg’s AI Journey
Greenfield vs. Brownfield Problem - AI coding assistance excels on new projects but shows diminishing returns on legacy systems. “Usage dropped really quickly once we moved back to brownfield.” Organizations must strategically choose where to deploy AI to avoid “vibe coding, where 2 engineers can create the tech debt of 50 engineers.”
Developer Motivation Over AI Maximization - Asking “what work do our developers not want to do?” drives higher adoption than mandating AI usage. Focus on eliminating tedious, undesired work creates sustainable engagement.
Agent-First for Context Assembly - In distributed systems, AI agents excel at assembling context rather than making decisions. Uplift agents generate patches with rationale and verification steps. Incident response agents aggregate overwhelming alerts and surface relevant context to reduce MTTR.
Bottom-Up Adoption Pattern - Individual contributors adopt AI tools at higher rates than leadership, inverting traditional enterprise rollout models. Marketing upward through IC success stories beats executive mandates.
Changing Engineering Economics - AI fundamentally alters the cost function of engineering. Traditional velocity metrics fail to capture the full value or risk profile of AI-augmented development.
Notable Content
- A Conversation about Developer Experience with Lei Zhang - Comprehensive interview on developer experience philosophy and infrastructure
- Meet the Team: AI Engineering at Bloomberg - Overview of Bloomberg’s AI engineering organization
- Durable DevOps Podcast Ep. 18 - Earlier discussion on DevOps and developer experience
About Bloomberg LP
Bloomberg is a major financial data, media, and technology company providing financial software tools, news, and analytics to professionals worldwide. The company employs over 9,000 engineers globally, maintains one of the largest private networks, and operates a massive JavaScript framework with tens of millions of lines of code across extensive internal libraries powering financial markets infrastructure.
Conference Appearance
Event: AI Engineering Code Summit 2025 Date: November 20, 2025 Time: 2:25 PM Session: What We Learned Deploying AI within Bloomberg’s Engineering Organization
Lei shared practical lessons from deploying AI tools and agent systems across Bloomberg’s large engineering organization. His talk covered the greenfield vs. brownfield problem, strategic AI deployment, organizational change management, uplift and incident response agents, training programs at scale, adoption patterns, and how AI changes the cost function of engineering in financial services environments.
Last Updated: November 24, 2025