Machine Learning Engineer

Yosef Zewdu

Building intelligent systems at the intersection of deep learning, generative AI, and neuro-symbolic AI. Focused on Agentic AI, computer vision, and applied machine learning research.

Featured Projects

A selection of research projects spanning generative AI, computer vision, and cognitive architectures.

Oracle Forge — Multi-Database Data Agent

A multi-database analytics agent built for the UC Berkeley DataAgentBench benchmark. Answers natural-language questions across PostgreSQL, MongoDB, SQLite, and DuckDB using MCP.

PythonLangGraphMCPPostgreSQLMongoDBOpenRouter
View Project →

Document Intelligence Refinery

An agentic document processing pipeline that automatically classifies documents and executes structured data extraction using a multi-stage LangGraph architecture.

PythonLangGraphDoclingPydanticFAISS
View Project →

OpenCog Hyperon: OpenPSI

A cognitive-affective architecture port to the MeTTa language within OpenCog Hyperon, modeling internal drives and emotions using symbolic hypergraphs.

OpenCogHyperonMeTTaSymbolic AI
View Project →

rPPG Remote Heart Rate Detection

A computer vision system that extracts heart rate from facial video using remote photoplethysmography (rPPG), enabling contactless vital sign monitoring.

PythonOpenCVPyTorchSignal ProcessingComputer Vision
View Project →

Medical Diagnosis Expert System

A MeTTa-based implementation of a classical rule-based system for clinical inference, demonstrating backward chaining for traceable diagnoses.

MeTTaPythonExpert SystemsNeurosymbolic AI
View Project →

Fraud Detection System

An end-to-end machine learning pipeline for detecting fraudulent transactions using ensemble methods and advanced feature engineering on imbalanced datasets.

PythonXGBoostScikit-learnSMOTEPandas
View Project →

Research Focus

Core areas driving my work at the intersection of intelligence, perception, and cognition.

🧠

Deep Learning

Neural network architectures and advanced training techniques for solving complex perception and reasoning tasks.

🎨

Generative AI

Diffusion models, GANs, and image synthesis — building systems that create, not just classify.

🔗

Neuro-symbolic AI

Combining neural learning with symbolic reasoning to bridge the gap between pattern recognition and structured thought.

🏗️

Cognitive Architectures

OpenCog, Hyperon, and frameworks for general intelligence — designing minds, not just models.

👁️

Computer Vision

Image processing, rPPG-based vital sign detection, and visual understanding from pixels to meaning.

From the Blog

Thoughts on AI research, deep learning, and the ideas I'm exploring.

May 2, 2024

Neuro-Symbolic AI: Bridging Deep Learning and Symbolic Reasoning with OpenCog Hyperon

Neuro-Symbolic AICognitive ArchitecturesOpenCogHyperonMeTTa
Read More →

March 15, 2024

Understanding Denoising Diffusion Probabilistic Models from Scratch

Diffusion ModelsGenerative AIDeep LearningPyTorch
Read More →