| Adversarial Prompt Armor | Benchmarking and recovering LLM robustness to adversarial prompt perturbations via inference-only Chain-of-Thought strategies. |
| Sentinel Cloud Vision | Real-Time Cloud-AI Streaming Engine: an end-to-end, cloud-native streaming platform that ingests high-frequency visual data for real-time AI inference, built to handle low-latency classification with a focus on MLOps and architectural scalability. |
| Open Model Foundry | Open-source multi-modal fine-tuning for LLMs and vision models, fully local, accelerating edge deployments. |
| Positional Encoding and Neural Rendering from Images | Implementation of Neural Radiance Fields in two stages. |
| QuickEmbed | Quick, local embedding inference for google/embeddinggemma-2 (740M, multimodal, Apache 2.0), with an optional cloud/GPU fallback and a terminal dashboard for watching queries and live compute usage on one screen. |
| Pluggable Agents | Stateful Multi-Agent System with LangGraph. |
