Anthony
Ngo
About
I’m an AI engineer with a Dr.-Ing. and around eight years building ML systems that have to hold up in production. I studied mechanical engineering at TU Darmstadt and earned my doctorate at the University of Stuttgart in collaboration with Bosch. Today I build agentic AI for capital markets at Deutsche Börse; before that, seven years at Bosch, from autonomous driving simulation and perception to LLM and agentic systems. This site is where I keep what I’m working on. For CV-level detail, ask the agent above.
Experience
- Building agentic AI for capital markets, with a strong emphasis on reliability, evaluation, and guardrails given the regulated domain.
- Developed synthetic data generation models for training and evaluating AI systems.
- Research and assessment of AI and generative techniques for high-fidelity synthetic data creation.
- Deployment and integration of generative models into simulation pipelines using neural rendering.
- Design and development of internal tools, such as automated model reporting and RAG-based LLM applications.
- Developed and validated CNN-based models for point cloud classification, benchmarked against traditional perception for detection, classification, and multi-object tracking.
- Built end-to-end pipelines for large-scale real and synthetic sensor data.
- Published peer-reviewed IEEE papers and contributed two patents.
Projects
Voyager↗
A self-hostable portfolio manager: broker CSV import, natural-language queries over holdings, German tax helpers, and a customizable widget dashboard.
Read-it-later
An AI read-it-later app that captures, summarizes, and resurfaces long-form reading with LLM-assisted organization.
Talking action figure
A DIY action figure running a local LLM on edge hardware, exploring on-device speech and personality.
Heimdall AI
An agentic RAG job-intelligence platform: a LangGraph pipeline with hybrid retrieval (keyword, semantic, RRF fusion), LLM query routing and relevance grading, and a RAGAS-inspired evaluation harness.