本服務為生技產業量身打造的 AI 資料自動化建模平台。使用者可上傳 CSV 格式的資料集與待分析的圖片集，系統將依據資料特徵自動對應單筆資料與其相關圖片。透過本介面，使用者可選擇用於建模的欄位並自定 AI 模型參數。完成運算後，即可獲得一套具備可解釋 AI 大數據模型，協助分析如製程中的推動與阻礙因子、實驗參數對結果的影響程度、預測良率，甚至辨識疾病跡象。平台亦提供預測工具，僅需輸入相對應資料，即可判斷新資料的適用性。
An AI-driven data automation and modeling platform tailored for the biotechnology industry. Users can upload CSV datasets along with corresponding image sets, and the system automatically maps each data entry to its relevant images based on data characteristics. Through an intuitive interface, users can select specific fields for model training and configure AI parameters as needed. Upon completion, the platform generates an explainable AI model that supports big data analysis—enabling users to identify process drivers and inhibitors, evaluate the impact of experimental variables, predict yield rates, and detect potential disease symptoms. A built-in prediction tool further allows users to validate new data entries for model applicability.
My Works
- Workflow
- UI flow
- UI design
- UX enhancement
- SA specifications
My Value
公司原本的 AI 模型訓練流程仰賴 AI 資料專家手動清洗資料、選擇欄位、調整參數，並透過程式指令反覆訓練模型，才能找出最佳的準確率。本平台的核心目標，是讓任何擁有資料的領域專家皆能自行訓練 AI 模型，並取得最終所需的預測結果。
我對本產品的貢獻在於，在 1 個月內從零開始與 AI 專家密切合作，經過多次設計與討論，打造出真正貼近建模流程的介面，讓即便沒有 AI 背景的使用者，也能透過直覺化的 GUI，完成原先需由 AI 專家手動操作的建模流程，同時自動生成可解釋的 AI 模型與預測工具，大幅簡化資料分析的門檻與操作複雜度。
The company’s original AI model training process relied heavily on data scientists manually cleaning datasets, selecting features, adjusting parameters, and running multiple iterations of scripted training to achieve optimal accuracy.
This platform was designed to empower domain experts—regardless of AI expertise—to independently train AI models and obtain the predictive outcomes they need.
My contribution to this product began from the ground up, collaborating closely with AI specialists through multiple rounds of discussion and design iterations in only 1 month. We developed a process-oriented solution that reflects the real-world model training workflow, while translating it into an intuitive GUI. As a result, even non-technical users can now build models previously requiring extensive manual effort, and automatically generate explainable AI and predictive tools—making data analysis more accessible and efficient.
