Binghamton University

Stanford University

University of Utah

Technologist, Investor and Entrepreneur

OpenResty Inc.

Seed Investor

Bay Area Fencing Club, distinguished as the nation's most rapidly expanding fencing establishment, has achieved a remarkable growth rate of 1600% within a mere two-year span. Our club stands apart in its commitment to not only teach the art of fencing but also to cultivate sportsmanship and character development among its members.

Bay Area Fencing Club

Investor

Seasoned Venture Capitalist Spearheading Seed-Stage Investments in Innovative Enterprise Software Startups

* General Partner: August 2023 – Present
* Venture Partner: February 2021 - August 2023

Foothill Ventures

General Partner

Building next-generation observability products.

AppDynamics

Engineering Leadership

Seed investor.

Pingcap is the creator of TiDB, which is an enterprise-grade distributed database that implements the MySQL protocol. For enterprises with MySQL scalability problems, Pingcap provides a no-code change solution. Pingap is the most successful open-source company in China. 

Github: https://github.com/pingcap/tidb

PingCAP

Advisor to following companies:

* Pingcap (https://pingcap.com)
* Imply (https://imply.io)
* StreamNative (https://streamnative.io)

Self-employed

Advisor

Love talking to entrepreneurs and help them in the areas of:

* Funding
* GTM strategy
* Technical directions

I invest in software companies in the following areas:

* Big Data
* Infrastructure
* Cloud-Native
* Open-source
* Observability
* Enterprise Software

Angel Investor

https://dashbase.io

(Acquired by Cisco AppDynamics)

Dashbase makes sense out of VoIP data + app logs for teams delivering voice, video, & chat
Dashbase is for quickly and proactively serving customers of unified communications, powered by VoIP & telemetry data

Dashbase

Co-founder and CTO

Worked on the next generation search infrastructure at Twitter.

Key accomplishments:

● User Search - Added realtime features, while reducing 99.9% latency from 1500ms down to 27ms. Reduced cluster size while sustaining corpus growth of over 200%. This was done in midst of a decision to double the cluster to handle increasing traffic, and resulted instead in halving the cluster size and still with 7x head room for traffic spikes.

● Social Filtering - Per query social filtering with early termination for user index. This greatly reduced cost while maintaining social relevance on the user index.

● Social Index - Incorporated the Twitter follow graph into the user index for supporting complex social queries, making filtering on large following lists possible. This removes the need for a separate social/follow graph cluster. This is similar to Facebook's Unicorn paper.

● Field attribution for relevance scoring - Enhanced Lucene's relevance model that provides attribution information from different parts of the query tree. This is necessary for layered reranking support.

● Realtime indexing core - Launched version 2 of the realtime indexing core that powers both tweet user indexes. This posting format removes the limitation that the previous posting encoding can have at most 255 positions. This technology was generalized into a memory codec for a generic Lucene application to handle in-memory indexing for realtime data.

● Search Root - Generic scatter-gather framework for parallelize work on homogeneous partitions of data in a distributed setting. This is used across different cluster types of Twitter's search corpuses.

Twitter

Staff Software Engineer & Tech Lead

Gave a guest lecture to the University of Utah's school of management on search technology and innovation.

Guest Lecturer

Contributed a chapter to Lucene In Action (2nd Edition)
on search technology behind LinkedIn.com:

http://www.manning.com/lucene/

Manning Publications Co.

Contributing Author

I oversaw the core technology behind LinkedIn’s pivotal products, including the Homepage and Search functionalities. Here are some highlights of my contributions:
* Leadership in Data Infrastructure: I led the development of the update stream data infrastructure that powered personalized and relevant activity streams on the LinkedIn Homepage. This system was designed for high performance and scalability, ensuring an efficient and responsive user experience.
* Co-Creation of LinkedIn Signal: I played a central role in conceptualizing and launching LinkedIn Signal, enhancing our content discovery tools.
* Architect of LinkedIn Faceted Search: I spearheaded the architecture and led the development of LinkedIn's Faceted Search, which was critical in refining search capabilities and improving user navigation.
* Development of Distributed Search Systems: I led a talented team to build a distributed search system that supported scalable social network-based people search. This initiative was a cornerstone in advancing LinkedIn’s next-generation search technologies.
* Repeated LinkedIn Hackday Winner: My innovative approach and commitment to excellence earned me multiple accolades at LinkedIn’s Hackday events, highlighting my creative problem-solving skills.

LinkedIn

Search Architect

http://www.plum.com/

Had a great deal of fun building social search solutions on a J2ee platform.
Built the search and browse system.
Built the API and RPC platform.
Built the event system.

Took owner ship of developing and architecting the back-end of plum.com.

plum.com (Now part of Nokia)

Software Architect - Search

Built the job search system from the ground up. Developed features such as job browse, job recency boost etc. Also built a benchmark suite for performance tuning.

The system is built on Apache Lucene.

Simply Hired

Search Consultant

Designed and built the social search solution for MyWeb 2.0: http://myweb2.search.yahoo.com.

MyWeb is a social search service from Yahoo! that allows users to share bookmarks. While working on this project, I solved interesting problems relating to duplicate removal, social network access control, scalability etc. and resulted in patents.

Yahoo!

Technical Yahoo! Sr.

Led a team of 3 engineers and developed the kernel of Entopia's search technology - K-Engine. This was a scalable distributed search solution that indexes social network, full-text and semantic contents that scaled to millions of documents. It was written in Java and deployed under the J2EE platform. The full text search aspect of the architecture was built on Apache Lucene.

Entopia Inc.

Software Architect

Co-created the Verity Parametric Search/Relational Taxonomy technology. This technology allows for real-time intersection of various types of result sets, e.g. full text, SQL-like selections and taxonomy browse sets. It powers applications to create various selections and taxonomies in order to make data catalogs truly browsable. Prototyped and productized this technology (which was a significant revenue generator for Verity). 

Led a team of 3 engineers and architected the Verity Federated Infrastructure. 

To provide an extensive framework to perform RPC calls to the backend server, developed and grew this into a platform where all Verity client APIs are built from. The framework is a RMI-flavored RPC model that contains TCP and RDBMS connection pooling, internationalization support and a transaction layer that communicates with a backend C++ server. This lays the foundation for Verity's hybrid Java/.NET to C++ architecture.

Autonomy (formerly Verity Inc.)

Implemented the billing solution for the game Everquest expansion pack - Ruins of Kunarc on the Portal Software's Infranet product using C++ on Unix.

Sony Online Entertainment

Senior Platform Engineer

Built ColdFusion and J2EE web applications for the Lockheed Postal Services Division.

Lockheed Martin

Senior Associate Software Engineer

Built 3D Graphics (OpenGL) drivers for image generators Harmony and Essemble. These image generators were primiarily used in flight and tank simulations.

John Wang