Thanh Tran

Online gaming industry is low tech and suffers from security and fairness issues including collusions, frauds and non-legitimate use of AI-assistance. Join my team of 200+ scientists, developers and designers to make online games more fair, safe and awesome.

Helped A5 Labs build tech & team.

Scaled Data/ML/Science team at Upwork from 5 to 50+. Successful helped Upwork go public & grow to $6+ billion company. Responsible for running the majority of core marketplace services and main partner in all data & ML based funnel optimization efforts. For details, see https://medium.com/upwork-datascience) and areas below: Search and Recommendation - Multi-sided matching: search and recommendation models to match clients and contractors & optimizing for marketplace constraints to ensure long-term health; requires learning-to-rank, learning-to-match, collaborative filtering, and unique methods for label generation and attribution over deep user funnels. Marketplace Optimization and Growth - Supply and demand modeling in two-/multi-sided marketplaces. - Optimal pricing: short-term marketplace optimality vs. long-term growth. - User modeling, user acquisition/retention/growth. Knowledge Graph - Build the de-facto standard conceptual model for online labor marketplaces; requires word/object embeddings, entity linking, assisted knowledge graph construction. - Knowledge graph applications: semantic query suggestion, semantic search and knowledge-based marketplace optimization. Data Science Infrastructure - End-to-end pipeline for standardized ML model training, scoring and validation; requires unifying stream and batch processing and scaling stream-based data processing for context-rich real-time model scoring. - Scalable platform for AB testing and product optimization: process and platform for more, better and faster experiments, bandit algorithms for online optimization.

2019: Helped A5 Labs build core AI team and develop solutions for player & game analytics and AI/bot detection to ensure security and fairness in online games. 2016: Helped Striiv built a scalable ETL+ML pipeline for handling large volumes of sensor data from wristband devices, and developed a ML solution to prove feasibility of detection of sleep apnea using PPG data. 2016: Led a stealth startup to build a chat-bot based social network for place and product recommendations using advanced tech for natural language conversations and searching over large-scale data. 2016-2019: Entrepreneur in Residence (Summer 2016 batch) and advisor for many StartX companies in topics such as building distributed teams, search technologies, Knowledge Graph, Big Data and AI.

Built and managed product & engineering teams across 3 locations. Patented and shipped Lyfeline Parenting Advice for new parents to receive on-demand chat-based consultation from pediatric experts. Shipped Lyfeline Milestones, which in 2016, has the highest number of App Store 5-stars reviews and downloads in the area of Child Development. Pioneered a ML-based solution that helps early detect autism (age of 14 months vs. average of 4 years) and together with Stanford researchers, developed a mobile therapy solution that can be administered by parents.

Together with researchers from KIT, built knowledge graph management and commercialized semantic search innovations.

Built and led one of the worldwide top research group in semantic search (measured in terms of publications and awards for tech innovations). Helped establish semantic search as a new research discipline and build research community around it via organization of the series of International Semantic Search Workshop.

Carried out research, developed software prototypes and managed research projects funded by the German Federal Government and the European Union.

Worked on managing and searching data (ontologies and textual documents) in the NCBO BioPortal, http://bioportal.bioontology.org/.

Developed J2EE-based customized integration and management softwares for DHL logistics (US branches).

Developed prototype and piloted one the first speech recognition/synthesis based automated job hotline in Germany.