I build production AI infrastructure and research how to make intelligent systems work where resources are scarce — constrained hardware, limited connectivity, non-technical users. My work spans LLM orchestration, RAG pipelines, and agentic workflows, with a security-first approach grounded in ethical hacking practice.
Pursuing an MPhil in Intelligent Computing Systems at Ashesi University, researching efficient AI in low-resource environments.
- The Question: How do we build intelligent systems that remain useful when compute is limited, connectivity is unreliable, and users aren't technical?
- The Stack: Lightweight model adaptation, retrieval systems optimized for constrained settings, semi-supervised learning for low-resource languages.
| Intelligent Systems | Security & Robustness | Systems Architecture |
|---|---|---|
| Agentic Orchestration | Adversarial Machine Learning | Distributed Systems Thinking |
| RAG Pipeline Optimization | Model Guardrailing & Safety | High-Performance FastAPI Backends |
| Fine-tuning & Embeddings | Secure SDLC Integration | Cloud-Native Scalability |
🏗️ Featured System: Meridian Policy Intelligence
A full-stack RAG application designed for high-precision corporate policy retrieval.
- Key Metric: Achieved 96% Citation Accuracy and 93.5% Groundedness.
- The Innovation: Implementation of a "Guardrail Prompt Template" and local CPU-based HuggingFace embeddings for cost-effective inference.
- Design: Editorial/Brutalist UI built with Vanilla HTML/CSS.
View Repository | View Live App
MPhil candidate at Ashesi University (Intelligent Computing Systems) and MSc candidate at Quantic (Software Engineering). My approach is defined by:
- Low-Resource AI: Researching lightweight model adaptation, semi-supervised ASR, and cross-lingual transfer for underrepresented languages.
- Security-First AI: Protecting intelligent systems against data poisoning and prompt injection — responsible AI and accessible AI are the same problem.
- Research-Driven Engineering: Applying academic rigor to production code, not just prototypes.
- Languages: Python (Expert), TypeScript, Nodejs, SQL.
- AI/ML: LangChain, PyTorch, HuggingFace, ChromaDB, Groq.
- DevOps/Backend: FastAPI, Docker, Kubernetes, AWS/Heroku.
- Security: Ethical Hacking, OWASP Top 10 for LLMs.