Building in the Open, Shipping for Everyone
We support open-source software development. Explore our public code repositories, fine-tuned model weights, and multi-agent systems.
Our Commitment to Open Weights & Frameworks
We believe that the technical developer community grows fastest when core engineering architectures, model weights, and system dependencies are shared transparently. Open-source codebases allow developers to inspect routing logic, verify security boundaries, and run custom model evaluations locally.
DBERT Labs actively maintains and contributes to public code repositories. We release specialized model weights optimized for edge devices, design modular orchestration packages, and collaborate on education-focused software packages.
Public Code & Models
DBERT_AI Custom Model
Our custom-tuned weights model trained on specialized data schemas and local execution scripts, designed to run smoothly on edge server configurations.
- Quantized checkpoints optimized for low VRAM targets (4-bit, 5-bit GGUF parameters)
- Excellent compliance for parsing workflows and returning structured JSON outputs
- Hosted and versioned directly on the official Ollama model registry page
Gayatri AI Repository Collaboration
An open-source educational framework providing schools and developers with offline-first, local multi-agent simulations.
- Modular Python libraries configuring communication channels between localized agents
- Automated assessment code setups to parse student prompt inputs
- Open-weights configurations to execute agent demonstrations without external API costs
Submit a Research Proposal
Are you interested in collaborating on academic briefs, contributing to open-source agent libraries, or testing local model configurations? Apply to DBERT Labs.
Submit Research Proposal