Jun Wan Email & Phone Number

Mountain View, CA, US

ML Platform and AI Infra for Google Ads safety and classification

San Francisco, CA, US

ML platform:1. Vector search platform to support LLM retrieval.2. Dolly team for LLM Dolly v1 and v2.3. Databricks Feature Store

Uber Tech Lead in Feature Store team of AI Infra Org:1. Build the FB company wide machine learning feature engineering platform. Include: Feature generation and extraction(serving) Infra for Batch, Realtime features. (2020-now)2. Build the Feature management platform for feature metadata, registration, metrics, lifecycle management, etc. (2020-now)3. Build.

Redmond, Washington, US

• Microsoft Office 365 Skype for Business User Data Management Team Design and implement the core cloud services to manage O365 Skype for business (SFB) user data. Including:
• Provision service: Super critical, solid, high perf and reliable service to provision worldwide 3000+ licensed users/sec into 12 Geo located SFB datacenters.
• Cross Forest migration service: Support big customer migrate the user data cross Geo located datacenters. The core service to support big multi-national company customers.
• Hosted Migration service: support enterprise customer migrate users from on premise Lync service into online.
• Meeting Migration service: Migrate the on premise Lync meetings into online to provide better performance and better Audio/Video quality.Lead the team for system design, technology and algorithm solution. Manage the.

Redmond, Washington, US

• Microsoft Lync Online Provision Service Team Tech lead of the team to design and implement services to provision the worldwide licensed users into cloud Lync datacenters. The service features include:
• Buildup 150+ sync stream to sync user data from Microsoft Online Directory Service into Lync online data center.
• Manage 700M+ users in the 12 Geo distributed cloud datacenters.
• Monitor user state change, license update and new tenants/users creation.
• Cooperate with 10+ other external Microsoft service. Build up a well-designed responsibility chain, error handling and retry mechanism.
• Super critical services, needs: high performance (3000+ users/sec), robust and high availability (customer escalation raises if service is down for mins), bug free (code bug will cause million users data issue).

Redmond, Washington, US

• Microsoft Communication Web Access Team, Microsoft China (2005 - 2009)The core developer that design and implement Web version of the complete Lync client. Includes:
• Asp.net web service to communicate with different Lync feature services, including: Authentication, contact list, IM/P, Conference, ACL, etc.
• Web client built by Ajax, HTML and JavaScript to implement the User login page, Contact List, Conversation window, Application sharing, User settings page, etc. (40,000 lines JavaScript and HTML code project)
• Outstanding Client framework design that separate the data channel with online service, user session state, and UI logic.

Redmond, Washington, US

• Microsoft Intelligent Troubleshooting service Team, Microsoft ChinaThe core developer that design and implement the online analysis service to trouble shoot Microsoft windows registry issues.
• A C# web service that consist of Frontend, SQL Database Backend.
• Build a giant database to store registry key diagnostic info.
• Creatively invent an algorithm to analyze customer registry issue in big data mining based on the Physics concept of “Entropy update”, which beat the original algorithm that created by a MSR mathematician by 35%.

Master'S Degree, Computer Science

Bachelor'S Degree, Physics