Peter (Chenda) Deng, Ph.D Email & Phone Number

San Francisco, California, US

Designing, developing, and implementing advanced LLM models for legal text analysis.Collaborating cross-functionally to integrate GenAI and NLP into legal document analysis and contract management workflows.Building legal data collection, processing, and integration pipelines for training and fine-tuning GenAI models.

Eden Prairie, MN, US

• Leveraged RAG and Langchain to build a vector database from knowledge articles for precise claim fraud detection, utilizing LLMs such as GPT and Llama2.
• Integrated Llama 2 (13b) and GPT seamlessly for HCC code extraction from medical charts, positioning us as leaders in coding accuracy and efficiency due to its superior performance over previous models
• Developed an decision support systems using an intelligence(NLP-based) framework for suggesting codes for clinical records. The project is on production and making millions of profits every year
• Researched and developed a product for processing dental clinical images using Computer Vision models to verify dental procedure codes

Nashville, Tennessee, US

• Utilizing computer vision models to identify different types of dental procedures from dental images
• Created an AI model for the segmentation of medical charts, resulting in significant annual profits
• Designed a BERT model for multi-label text classification of PubMed articles
• Developed a Named Entity Recognition (NER) model to detect encounter dates in medical charts

Palo Alto, CA, US

• Develop a question-answering system capable of processing both structured and unstructured datasets.
• Implement various Natural Language Understanding (NLU) models for Intent Classification and Named Entity Recognition.
• Utilize advanced NLU and Natural Language Inference (NLI) models, including BERT, ALBERT, and RoBERTa, for in-depth analysis of Electronic Health Record (EHR) data.
• Employ Natural Language Generation (NLG) techniques to create relevant and contextually appropriate questions.
• Engage with publicly available healthcare datasets, such as those from ClinicalTrials.gov and PubMed, for research and application development.

Palo Alto, CA, US

• Enhancing evidence retrieval and Recognizing Textual Entailment (RTE) for medical claims
• Improving translation capabilities for medical records
• Conducting data analysis and designing backend systems for structured healthcare data

Fort Collins, CO, US

• Implement Parallel Programming (MPI) to integrate a 3-D model (Modflow) and a 2-D model (Daycent) for coupled simulations.
• Apply Machine Learning techniques to forecast Groundwater levels.

Vantaa, FI

• Lead scientist on a traffic data analytics and modeling project, successfully developing a prototype computer learning model using traffic and Road Weather Information System (RWIS) station data to provide enhanced.
• Processed, organized, and stored over 500GB of traffic and RWIS data from company sources and third-party data providers
• Mapped the data using QGIS in Python
• Constructed learning models using Python scikit-learn (Artificial Neural Networks, Support Vector Regression, Random Forest) and TensorFlow (Long-Short Term Memory/LSTM)
• Worked in an Amazon Web Services (AWS) Elastic Compute Cloud (EC2) virtual machine with Graphical Processing Unit (GPU) architecture
• Created a novel method to estimate traffic flow

Doctor Of Philosophy - Phd, Engineering

Master Of Science - Ms, Engineering

Bachelors Of Science In Civil And Environmental Engineering, Soil And Water Resourse