Mountain View, California, United States
2025 focus on Agentic AI and Automation, create the new Agentic AI commercial model for AI agents-based solutions that create new revenue streams and enthusiastic customers. Lead 2024 Google AI & Compute (ROFL) projects that enriched every user's initial experience to support product AI strategy. Delivered measurable results to reduce lead time and reduce capital expenditure via improving forecast accuracy and optimize cost through projects in Cloud AI and Compute, Core ML, Demand and Supply chain pipelines. Passionate about taking product design from 0 to 1 and from 1 to 10. Alongside my full-time role, I'm passionate about accelerating early-stage startups in UX design, research, and 0-to-1 product development. Specializing in AI, Cloud Infrastructure and Fin-tech. Outside of work, I enjoy sailing, tennis, and reading, which refresh my body and mind. After graduating from Master's degree in Computer Science Engineering and Design (HCI focus) from Harvard University, I applied design thinking, user research, prototyping, and evaluation methods to various Machine Learning domains and design contexts. I also have a diverse educational background in statistics, interactive media design, and economics, which enables me to approach design challenges from multiple perspectives and leverage data-driven insights.
AWS “It is still day 1”
· Using a new era of experience - Agent Experience (AX) Design to re-frame Agent AI pricing model through agentforce, which will create new revenue streams and enthusiastic customers. · Pairing product roadmap, from 0-1, creating pricing estimate agent, smart usage alert agent and capacity planning agent to sizing right product for enterprise customers, equip Salesforce AEs with upsell power and boost ecosystem capabilities.. · Optimize AI agents capabilities through MCP via multiple agents as a suite to fulfill enterprise and SMBs needs, by analyzing historical usage and monitoring current API metrics, customers are able to fully aware and detect usage change and make the best move in fast changing business world. · Bringing AI agents into customer touchpoints, opening up entry points via slack, ROI calculation tool and future A2A framework possibilities. Mapping E2E procedure to boost sales account executive power to bridging business value insights, historical usage and future capacity planning in one loop. · Leverage AI-driven conversational experience across core AI CRM, mulesoft and data cloud, generating 2025 vision interaction design, which win COO and product lead acknowledgement in real-time usage and billing. Foster digital wallet experience across multiple product lines, creating E2E one source of truth for users. · Shaping customer vision into real mental models which serve customer needs aligned with Agentforce, Data Cloud, and Customer 360 App, crafting vision and JTBD.
· Leading 2024 key ML projects in Cloud AI and Compute, Core ML, TI demand and supply chain(2023), which enriched user experience by unifying multiple tools into a single pane CUJs, supporting product and engineering AI strategy. · Enhanced the product, business and engineering solutions with UX support in the pipeline that predicts, plans, and deploys networking, datacenter infrastructure requirements to satisfy compute and storage demand, shortening lead time to deploy machines, reducing capital expenditure via forecast accuracy, further support changing AI roadmaps. · Leading Fleet Management UX – MVP launched in 3 months which helps effectively manage and maintain optimum fleet Inventory levels across all product areas in Google– Product + UX winning Quarterly Performance Award (Q3 2023). · Collaborated with the cross-functional (engineering, product, business), landing UX mock across platform modules, rapid prototype to help simulate scenario creation, which further translated into AI ecosystem that optimizes and automates the Google datacenter resource planning of Compute, Storage, and ML(GPU, TPU, etc.) resources. · Facilitated UX workshop to mapping E2E flow that spans demand-supply planning, forecast and fulfillment to enable users across leadership, business, marketing to make informed strategic decisions. · Within the information architectural framework, swiftly outline various AI UX presumptions to test hypotheses among users. Gather feedback and provide constructive input to engineering and product teams to facilitate the development of mutually beneficial solutions – which reduced * request turnaround time and saved * engineer-hours. · Compile a summary of relevant projects 23-24 to refine UX focus-AI power strategy and vision, which further optimize resource allocation, machine delivery and capacity distribution. · Illustrating UX success metrics, aligning leadership to drive strategic initiatives to deliver top-notch features. (vendor)
Partner effectively with local Small & Medium Business, UX researchers, product, marketing to understand user pain-points. Conceptualize a broad spectrum of ideas and then narrow based on experimentation and customer research. Designed and shipped multiple features based on a lean UX build-measure-learn process in QuickBooks-Online-Advanced, QB-money/Capital, launch received 975 customer votes ratings with an average of 4.75/5 stars.
· Meta Orion wearable device, Ray-Ban glasses (funded by Meta, co-op with Harvard faculty side project). User research on smart glass: Multimodal machine learning model scan and HCI research, narrowing the scope to investigating different dimension reduction and attention mechanisms, listing strengths and weaknesses of different approaches. (both theoretical and applied on a multimodal dataset) using python library. · Meta (FRL) Research: EMG wearable device – use a built-in sensor to collect data, perform data exploration, feature extraction, model training, testing and deployment in edge impulse. Wristband technology that uses muscle signals as a form of input can facilitate more inclusive human-computer interactions (HCI) for people with a wide range of neuromotor abilities. HCI research for wearable devices (connect sensor-tile hardware-software and enterprise applications). · FDA AI Search and Conversation Design: Dive deep with AI in UX — Using ML model and NLP, which support user search in open FDA Drug label library. The CUJ model traces behaviors of machine learning models back to their training data, and influence functions aim to predict how the trained model would change if a specific training example were added to the training set. Illustrate pain-point that block the applicability to large-scale neural nets; apparent inaccuracy of the results, and the difficulty of computing. Further, UX approaches to scaling influence estimation to large language models and show data visualization insights of patterns of generalization.