“I can’t guarantee you a win — but I can engineer systems that make failure mathematically expensive, bounded, and observable.”
Chief Engineer @ Graceman
I’m David Grace (
davidgracemann) — Chief Engineer at Graceman and the owner/maintainer of the Graceman organization. This GitHub profile is my personal research space and is standalone from any organizational work done at Graceman
Research Masters in Computer & Systems Engineering @ Technische Universität Ilmenau
| Domains Of Focus | What I do | Status / Notes |
|---|---|---|
| Computer & Systems Engineering (Research) | Academic research aligned with my Master’s program at TU Ilmenau. | Active 🟢 ( Academia ) |
| AI (Foundational + Applied) | Proof-of-designs, implementations, and applied research with rigorous evaluation where possible. | Active 🟢 ( Core Focus Domain ) |
| Quant + High-Performance Computing | Performance-first engineering, low-latency systems, numerical methods, execution-grade infrastructure . | Active 🟢 ( Honestly I need to eat and pay rent ) |
| Legacy Fintech / BankOps | Maintenance-only support for older open-source contributions from earlier work. | Maintenance-only 🟡 ( For Old Time's Sake ) |
I have pivoted to personal research. No new Fintech/BankOps feature work is planned (maintenance-only).
Focusing on transitioning existing "BankOps" systems from standard microservices to formal, event-sourced architectures with enterprise-grade rigor.
| Project | Current State | Hardening Objective |
|---|---|---|
| FlossPay | Payment Aggregator | Implementing L4 circuit-breaking and DLQ patterns for PCI-DSS compliant fault tolerance. |
| Flossx83 | ISO 8583 Simulator | Integrating HSM-grade AES256 tokenization and RLT Fraud Detection via ML microservices. |
| Recon-Engine | Distributed Recon | Upgrading to an event-sourced architecture with hash-collision detection for 99.5% auto-recon. |
Advancing domain-specific AI models through fine-tuning and mathematical primitive research.
| POC Name | Research Focus | Technical Goal |
|---|---|---|
| IRSIE | SLM for Tax Jurisprudence | Fine-tuning a 12-layer transformer on 2.8B tokens with RLHF-tuned deterministic validation. |
| Qaml | Mathematical Transformers | Training a 65M-parameter model from first principles for sub-50ms suspicious pattern detection. |
Pioneering projects at the intersection of Multi-Agent Systems (MAS) and Systems Engineering for your upcoming research at TU Ilmenau.
| Project Title | Category | Strategic Objective |
|---|---|---|
| Preemptive Threat Agent | Applied AI #2 | Architecting an autonomous agentic framework for defense simulations via Stochastic Manifold Analysis. |
| Graceman Graph-Map | Theoretical CS #1 | Formal verification of self-healing software states using TLA+ and recursive graph mapping. |
Would you like me to generate a detailed technical breakdown for the "Preemptive Threat Agent" to use as a standalone project highlight on your profile?
- Email: [email protected]
- Org: github.com/gracemann365