Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I am Tran Huu Duc (Peter) Phan, an AI Engineer major in multimodal deep learning, neural signal processing, and Generative AI. I have proven expertise in building reliable machine learning pipelines, including fNIRS and EEG data fusion to improve decoding accuracy in noisy environments, and I enjoy building end-to-end Retrieval-Augmented Generation systems with Llama 3.2 and hybrid search to deliver document-based AI solutions. In addition to research, I have hands-on experience deploying fraud-detection systems (e.g., graph neural nets for fake review rings) and multi-modal brain signal fusion projects. I thrive in collaborative environments, value clear evaluation frameworks, and am eager to apply my skills to new challenges in AI engineering.…I am Tran Huu Duc (Peter) Phan, an AI Engineer major in multimodal deep learning, neural signal processing, and Generative AI. I have proven expertise in building reliable machine learning pipelines, including fNIRS and EEG data fusion to improve decoding accuracy in noisy environments, and I enjoy building end-to-end Retrieval-Augmented Generation systems with Llama 3.2 and hybrid search to deliver document-based AI solutions. In addition to research, I have hands-on experience deploying fraud-detection systems (e.g., graph neural nets for fake review rings) and multi-modal brain signal fusion projects. I thrive in collaborative environments, value clear evaluation frameworks, and am eager to apply my skills to new challenges in AI engineering.

Tran Huu Duc (Peter) Phan

AI Engineer, Data Scientist, Full Stack Developer, +2





I am Tran Huu Duc (Peter) Phan, an AI Engineer major in multimodal deep learning, neural signal processing, and Generative AI. I have proven expertise in building reliable machine learning pipelines, including fNIRS and EEG data fusion to improve decoding accuracy in noisy environments, and I enjoy building end-to-end Retrieval-Augmented Generation systems with Llama 3.2 and hybrid search to deliver document-based AI solutions. In addition to research, I have hands-on experience deploying fraud-detection systems (e.g., graph neural nets for fake review rings) and multi-modal brain signal fusion projects. I thrive in collaborative environments, value clear evaluation frameworks, and am eager to apply my skills to new challenges in AI engineering.…I am Tran Huu Duc (Peter) Phan, an AI Engineer major in multimodal deep learning, neural signal processing, and Generative AI. I have proven expertise in building reliable machine learning pipelines, including fNIRS and EEG data fusion to improve decoding accuracy in noisy environments, and I enjoy building end-to-end Retrieval-Augmented Generation systems with Llama 3.2 and hybrid search to deliver document-based AI solutions. In addition to research, I have hands-on experience deploying fraud-detection systems (e.g., graph neural nets for fake review rings) and multi-modal brain signal fusion projects. I thrive in collaborative environments, value clear evaluation frameworks, and am eager to apply my skills to new challenges in AI engineering.

Available to hire

I am Tran Huu Duc (Peter) Phan, an AI Engineer major in multimodal deep learning, neural signal processing, and Generative AI. I have proven expertise in building reliable machine learning pipelines, including fNIRS and EEG data fusion to improve decoding accuracy in noisy environments, and I enjoy building end-to-end Retrieval-Augmented Generation systems with Llama 3.2 and hybrid search to deliver document-based AI solutions.

In addition to research, I have hands-on experience deploying fraud-detection systems (e.g., graph neural nets for fake review rings) and multi-modal brain signal fusion projects. I thrive in collaborative environments, value clear evaluation frameworks, and am eager to apply my skills to new challenges in AI engineering.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Language

English

Fluent

Work Experience

Fraud Detection ML Engineer at Amazon

July 1, 2015 - November 1, 2015

Developed an end-to-end AI pipeline using Graph Neural Networks to identify fake review rings, achieving high accuracy (0.85 Test AUC) on large-scale datasets. Built a Heterogeneous Graph Transformer to map complex relationships between users and products. Combined AI-generated text features with social network patterns to detect coordinated fraud. Improved performance on rare fraud cases by using Contrastive Learning and synthetic data techniques to balance highly uneven datasets.

ML Engineer - End-to-End Hybrid RAG Pipeline at Independent project

November 1, 2025 - January 1, 2026

Developed a complete Retrieval-Augmented Generation system using Python to query complex PDF research papers. Built a Hybrid Search engine (FAISS + BM25) with RRF to ensure the most relevant data is retrieved. Integrated RAGAS Evaluation to measure accuracy and Gradio to create a user-friendly chat interface with source citations. Optimized data retrieval by breaking down large PDF documents into smaller, searchable chunks to improve the speed and relevance of AI responses. Reduced hallucinations and improved factual accuracy by forcing the model to provide direct citations from the original research papers.

Multi-Modal Brain Signal Fusion Researcher at Independent project

July 1, 2025 - Present

Built a data cleaning pipeline for fNIRS and EEG signals using MNE-Python to remove noise and motion errors. Combined EEG and fNIRS data into a single system to improve the accuracy of decoding imagined speech compared to using only one signal type. Wrote a detailed research paper covering signal selection, AI feature extraction, and future ways to improve the technology.