Yaozhang Pan

Serve Bamboo School as Chief EdTech Officer (CETO) , Taking primary responsibility for the development of the school-led AI self-learning platform (including the learning engine, cognitive models, and data middleware/platform), establishing an intelligent closed loop of “teaching–learning–practice,” promoting technology-enabled growth for both teachers and students, and leading the transformation and application of industry–academia–research.

LearnScape is an AI-powered, gamified learning platform designed to transform how children engage with language education. Tailored for K12 learners in Southeast Asia, LearnScape blends emotional intelligence, adaptive learning, and storytelling into an immersive learning world. At the heart of LearnScape is the AI Buddy — a personalized learning companion that grows with each student, offering contextual feedback, encouragement, and interactive quests. Through memory-based review loops, open-world exploration, and multilingual support, LearnScape ensures every child not only learns, but enjoys the journey. With a smart content creation toolkit for teachers and parents, LearnScape bridges the gap between motivation and mastery, making it easy to personalize learning at scale. Our vision: To create an emotionally aware, personalized learning universe where every child feels seen, heard, and empowered.

Heading the overall Data Office, which includes data engineering, data analytics, and data science functions. In the Tokopedia data office, we see ourselves as an integrated team in solving Tokopedia and GoTo problems. Our #OneDataOffice vision aims for democratizing Data, knowledge, and AI capabilities throughout GoTo and Tokopedia scalably.

Roadmap planning and execution  • Plan technology roadmap, communicate across different functions, • Identify business-wide problems and executed critical data science projects • Our data science projects cover all major AI areas: Nature Language Processing, Computer Vision, Image Processing, Knowledge Graph, Deep Learning Manpower planning, talent sourcing • Building a top data science team from scratch up to 100+ full-time data scientists and 20+ data science interns. • Initiated collaborations with Universities and data science communities, co-organized several meetup and machine learning competitions, attracted 200-300 talents each time.

Leading the data science team with 30+ team members for the E-commerce Business of SEA we're hiring, feel free to reach out!

● Identify business-wide problems and lead junior data scientists in solving them using a combination of Big-data, Machine Learning (ML), and/or Operations Research (OR) techniques. ● Initiate, plan, and execute multiple projects. Work across different functions in the business. ● Communicate problem formulation, solution, analyses and insights to team members and stakeholders. Selected Projects: Grabshare One of the milestones I achieved in Grab. I was in charge of the matching algorithm design and worked as data owner in this top priority project. The GrabShare service pairs just two passenger bookings with similar trip routes within a single trip. We use optimization technologies to make the GrabShare's matching algorithm ensures passengers get to their destinations in the shortest possible time. POI (place of interest) Ranking Designed and deployed a search ranking algorithm, which increased top 1 hit accuracy from 30% to 70%, top 5 hit accuracy from 70% to 97%.

Work as a research scientist who is exploring and developing all kinds of new technologies and innovations, specially focus on developing cloud analytics stack and applications of cloud analytics. Some of my works at Greenwave Systems, with analytics being the most recent: Data Analytics - Hadoop, Mahout, Hive, Impala, Cassandra, Spark, Shark - Simulation and cost optimization of home heating and cooling systems. - Recommendation system for advocating energy monitoring and management in homes. Projects: Smart Heating: This project relates to cost optimization for premises that are equipped with electrical heating or cooling systems for regulating indoor temperatures. The cost of electricity varies by the hour (or other units of time) and such price schedules are published by the utility company a day ahead, or at least prior to when our cost optimization calculation is performed. We explore how two variables can affect the costs of heating/cooling: target temperate and start-time of heating/cooling, and present various scenarios to the occupant to optimize the heat pump setting that gain a good balance between energy saving and comfort. Analytic stacks that used in this project including Hadoop, impala, hive etc. Solar Panel: This project focus on fault detection of solar panel. We keep monitoring the power output of solar panel and based on the output data to analyze and identify the situation and fault of solar panel. For example, snow lowers power output from the solar panel. Individual panels can fail, wiring can break, inverter can fail – all leading to lowered or zero output of solar panel. If solar panel output low and weather forecast snow, it probably snow. If the weather is good but the power is low, the problem is probably with wiring. If the power consistently low, then maybe 1 or more panels failed. Analytic stacks that used in this project including Hadoop, hive etc.

Research on EEG signal processing, machine learning, brain-computer interface etc. Worked on projects helping stroke and epilepsy patients: iSLEEP: through a novel brainwave analysis method, developing a system which is able to continuously play back the most suitable music that can help the user enter sleep, in real time, according to the person’s brain status. The system is able to help people enter sleep quickly and easily, therefore improving their quality of sleep. I am the technical leader of this project and focused on the brainwave-music response modeling; MATLAB code implementation; inventive steps development; system performance validation; technical documentation, etc. iMUSIC: providing a brand new music sorting mechanism which automatically and intuitively selects a piece of music or ranks a music playlist according to the listener’s emotional information from brainwaves, derived from a real-time Brain Computer Interface (BCI). I am the technical leader of this project and focused on designing the neural signal processing methods, Matlab implementation of signal processing codes, technical documentation, and experimental analysis, etc. SEIZE: Automated video-EEG analytic system in seizure detection in the Epilepsy Monitoring Unit, 01 Feb 2010 - 30 Sep 2012. As an expert in seizure detection using EEG signal, I focused on analyzing long-term EEG signal and build the machine learning algorithm to automatically detect both general and focal seizures and localize the focal seizures. ArtsBCI: Advanced rehabilitation therapy for stroke based on Brain-Computer Interface, 01 Feb 2010 - 30 Sep 2012. I worked on fMRI image investigations based on pattern recognition methods to evaluate the effect of BCI based rehabilitation for stroke patients.