Dongchen Wang

Key contributions: ---Built and managed Retrieval-Augmented Generation (RAG) chatbot pipelines for patent search, resume screening, improving information retrieval for R&D and HR teams. ---Designed text-to-SQL AI assistants enabling chemists to query large database of multilingual experimental records. ---Worked closely with HR and IT teams to deliver information extraction chatbots across subsidiaries. ---Developed evaluation frameworks using open-source LLMs and SQL-based feedback systems for continuous improvement. ---Responsible for developing AI solutions and managing project timelines; responsibilities include coordinating with internal departments such as R&D, HR, and IT, and working with external vendors. Tasks include ensuring stakeholder alignment, tracking progress, and supporting solution delivery through clear and structured communication across teams. Tools & Tech: Streamlit, LangChain, Ollama, SQL Server, PostgreSQL, Docker, FastAPI, Llama, OpenAI, DeepSeek, and more.

Scalp Diagnosis Project: ---Collaboration: Partnered with domain experts to create an expert-level scalp diagnosis application. ---Requirements Analysis: Communicated with clients and domain experts to define product requirements and tailored solutions. ---Data Annotation: Recruited and managed annotators, ensuring high-quality data for model training. ---Model Development: Built deep learning image classification models with over 85% accuracy using PyTorch, OpenCV, YOLO, Ray Tune, Matplotlib and others. Face Diagnosis Project: ---Research and Problem Formulation: Conducted in-depth research on medical and aesthetics studies to formulate precise definitions of facial issues. ---Image Segmentation: Reviewed and implemented a range of image segmentation techniques, from classical methods to SAM, to identify fine-grained facial features, including acne. ---Pipeline Development: Designed and constructed classic image segmentation pipelines, integrating them into production systems for real-time analysis. ---Advanced Image Processing: Utilized and explored advanced techniques such as image preprocessing, color space analysis, and key point detection extensively throughout the project to enhance facial analysis accuracy.

Master of Technology in Intelligent Systems CAP: 4.13 Toy Projects: ---Designed a modal that classifies patients' requests into different levels of emergency through NLP and sentiment analysis. ---Designed a model to sense the emotion in patients' request voice recordings. ---Designed a model to reason and identify the pedestrian light signal (green, red, blinking) in video recordings. ---Designed a model to map social media sentiments to targeted entity and produced analytic scores based on predefined/dynamic aspects.

---Introduced and led the transformation of the reading program to the E-reading program. The E-reading program has encouraged students to enjoy reading during Covid 19 period. ---Level Representative: Communicated and coordinated with level teachers and department heads on teaching and learning-related activities. Manage curriculum schedule and prepare teaching resources. ---Group Lead for programs/projects: Plan research content, oversee project timeline, map out tasks and delegate tasks to group members, and provide help and support.

---Involved in designing a prototype website for a new funding program. ---Analyzed and researched on the trend of past Singapore government funding on building energy efficient research projects. • Learnt the administration of funding/grant. ---Completed literature reviews on construction materials, building modeling and simulation, and energy-efficient building technologies.

---Assisted senior engineers in the drafting of a preliminary Environmental Assessment List. ---Learnt information gathering for drafting Environmental Impact Assessment (EIA) reports through site visits. ---Learnt the drafting of EIA reports through reading the archives and participating in the preparation of EIA reports.

---Written articles for U-Weekly magazine. The articles include recipes, supernatural stories, fun places, good foods, activities and special reports on Mother’s Day, Ester etc. ---Collaborated with the designing department and the advertisers to create and design advertisement for the advertisers.