Nguyen Quoc Anh - AI Engineer
Biography
Nguyen Quoc Anh earned his Bachelor's in Economics and Finance at the Business School, RMIT Vietnam. He served as an AI/ML Researcher and Founder of The Neurone Lab. Currently, he is an AI Engineer at Hitachi Digital Services, specializing in Computer Vision and Generative AI applications. Anh excels in interdisciplinary studies, merging traditional finance with data science. He pioneered predictive models using Chaos Theory to enhance forecasting accuracy, advancing algorithmic trading through decentralized AI. Proficient in SQL, R, and Python, he collaborates with Ph.D. researchers from prestigious EU institutions on societal time series.
Education
Bachelor of Business in Economics and Finance, 2024
Royal Melbourne Institute of Technology
Work Experience
- Project Virtual Planner (POC) - BackEnd GenAI Engineer
- Developed LLM-based (Llama3.3-70B, Gwen2.5-70B, Azure OpenAI) multi-agent systems for high-speed rail management, focusing on information retrieval, penalty optimization, and plan scheduling
- Applied LangChain/LangGraph frameworks to construct a Text-to-SQL agent from scratch, managed ETL transformation on PostgreDB, deployed weekly on FastAPI for internal testing
- Programmed automatic violation classification utilizing similarity search via in-context learning and FAISS index; backtested RAG, CPAL chain, and CoT prompting techniques
- Designed professional system prompts following industry's protocols, reducing agents' hallucination thereby saving query cost per output
- In control of the multi-agent pipeline, defining and importing functions for optimal chatbot experience
- Engaged in conceptual and high-level design (confidential) of the agentic system
- Project Nestle HoloLens (Deploy) - Computer Vision Engineer
- Trained, fine-tuned, and tested Yolov8-11n models on HoloLens v2 for real-time object detection and stage completion in factory manufacturing
- Wrapped post-train models in ONNX format for HoloLens deployment, managed quality control and testing
- Experimented with diverse augmentation parameters
- Project Station Finder (POC) - Technical Advisor
- Assisted in Text-to-SQL agent development, advising on system prompts and database design
- Conducted user study (n=1000) on unreleased smart devices, instructing users through 50+ protocols
- Engaged in CRM training and maintained professional customer service in daily F2F contact
- Set up working environment from scratch per vendor requirements, ensuring data privacy
- Developed deep Neural Network models for S&P 500 stocks on quant exchange
- Optimized metrics with Phase Space Reconstruction and Attention Mechanisms
- Managed ETL processes for 2M irregular observations from PhysioNet and YFinance using Python
- Built cryptographic algorithms and Federated Environments for IoT cybersecurity; peer-reviewed 43 PhD papers
Research Interests
Projects
CLAM: A Stacked CNN-LSTM-AM Model for Stock Trend Prediction
- Developed a synthetic model with stacked layers of Conv1D, LSTM, and Attention Mechanisms for multi-step stock trend forecasting.
- Fine-tuned CLAM through 48 hyperparameters at 90:10 split, optimized training with EarlyStopping and ReduceLROnPlateau callbacks.
- Improved MAE and RMSE by 90%, capturing 75% of out-sample stock trends with flash crashes, outperforming LSTM and CNN.
CryptMAGE: Applied Vision Transformer for Intraday Cryptocurrency Time Series Pattern Recognition
- Fetched 2-year hourly crypto data, transformed intervals into images, embedded static info as metadata for optimal training.
- Ensured label balance, engineered MA and RSI indicators, developed correlation heatmaps between OHLCV and datasets.
- Reduced intercorrelation by 53%, achieved 95% accuracy, 91% AUROC, and 90% AUPRC with Swin Transformer over ViT and DeiT.
PSR-NODE: A Phase Space Reconstructed Neural ODE Model For Financial Forecasting
- Programmed and fine-tuned NODE for time series regression, optimized with embeddings and time delays across 32,000 observations.
- Leveraged Takens’s Theorem, designed 3D phase space, captured cyclical and long-term dependencies across stock sectors.
- Applied TimeSeriesSplit and 5-fold validation, improved MAE and RMSE by 70%, outperforming LSTM, Transformer, SVR.
CNN-BiLSTM-GRU Model for Stock Price Forecasting
- Developed a hybrid deep learning model, optimized with tailored embeddings and time delays for PSR on 20-year data.
- Forecasted stock prices for Target (TGT), Amazon (AMZN), and Walmart (WMT), reducing MAE by 37.2% and RMSE by 35.1%.
- Presented at Digital3 Conference, RMIT Experience Day, and RMIT Showcase: Impact in 2023.
Publications
In Intelligenza Artificiale, Q2, 1(1):1-17. DOI: 10.1177/17248035251322877
In Proceedings of The 31st Pacific Asian Conference on Information Systems (PACIS 2024), CORE A, 1(1):1-17.
In Proceedings of The 21st International Conference on Modeling Decisions for Artificial Intelligence (MDAI 2024), CORE B, 1(14986):1-14. DOI: 10.1007/978-3-031-68208-7_20
SSRN Electronic Journal, 1(1):1-16. DOI: 10.2139/ssrn.4729648
In Proceedings of The 9th International Conference for Young Researchers in Economics and Business (ICYREB 2023), UEH Publishing House, ISBN: 978-604-346-250-0, Vol. 2, pp. 83-94. DOI: 10.5281/zenodo.11081926
YouTube, Sponsored by Spiderum Official, 900K Subscribers Channel.
Awards and Honors
- 2025, Spring - Invited Peer Reviewer, IEEE Transactions on Industrial Informatics
- 2024, Winter - Guest Speaker, Vietnam Investment Summit, Bloomberg Businessweek Vietnam
- 2024, Winter - Presenter, Vietnam Economist Annual Meeting, EIU
- 2024, Summer - Presenter, Pacific Asia Conference on Information Systems (PACIS 2024), HCMC
- 2024, Summer - Invited Peer Reviewer, Computers & Electrical Engineering, Humanities & Social Sciences Communications, Engineering Applications of Artificial Intelligence
- 2024, Spring - Invited Peer Reviewer, ICPR 2024, Annals of Operation Research, IEEE Access
- 2024, Spring - Invited Speaker, ICEFMR-2024 Conference, Japan
- 2024, Spring - RMIT Showcase: Impact, Online School Showcase
- 2023, Winter - Project Love Meter Winner, RMIT Showcase, Hall of Fame, RMIT Vietnam
- 2023, Fall - Presenter, ICYREB International Conference, UEH
- 2023, Fall - Presenter, Digital3 International Conference, RMIT
Featured Media
RMIT Hall of Fame 2023
