Kossiso Udodi

timber. Ollama for classical ML models. AOT compiler that turns XGBoost, LightGBM, scikit-learn, CatBoost & ONNX models into native C99 inference code. One command to load, one command to serve. 336x faster than Python inference.

train_grpo.py. GRPO Training Script for Qwen Model on GSM8K Dataset. This script trains a Qwen model using the GRPO (Generalized Reinforcement Policy Optimization) method on the GSM8K (Generalized Math 8K) dataset. The script leverages transformers, PEFT (Parameter-Efficient Fine-Tuning), and TRL (Transformers Reinforcement Learning) libraries.

economic-policy-simulator. This project combines a large-language-model (LLM) for natural language policy input, an agent-based computational economics (ACE) core for simulation, and real-time fiscal feedback loops.

Gemma-3n-local-training. A lightweight, GPU-focused framework to run inference and LoRA fine-tuning on Google’s Gemma 3n family (`1.1B`, `2B`). Designed for small-scale deployments such as chatbots, assistants, or domain-specific Q&A systems.

vxdf. VXDF (Vector eXchange Data Format) is an AI-native container for text, metadata and vector embeddings—portable, indexable and compressed.

vxdf-platform. TypeScript

maxa-ai. Maxa is an AI assistant that maintains persistent memory and theory of mind capabilities, enabling more natural and context-aware interactions over time.

HMT-Zero. Continuous AI Consciousness Architecture

Loan-coursework. Jupyter Notebook

ruth. Google Maps for your codebase — interactive dependency graph, landmark detection, code preview, real-time topology updates