Mboh Bless Pearl Nchongboh

Working and researching on projects that need integration and improvements in their use of ML models and also developing applications around the ML models

• ▸Reduced model inference latency by 60%
• ▸LLMs
• ▸Computer Vision
• ▸Led multiple teams and projects

I designed and built the first version of an automated AI-driven security report generation system for Android application scans. My work focused on extracting and structuring relevant features from large, unstructured scan files to make them usable by language models and multi-agent analysis pipelines. I evaluated multiple LLM approaches, including fine-tuning, RAG and advanced prompting methods, to determine the most reliable strategy for vulnerability analysis. To handle heavy workloads under limited compute, I implemented a queue-based bulk-processing system optimized for open-source models running on constrained, self-hosted infrastructure.

• ▸Delivered the first production-ready AI security report generator for Android application scans
• ▸Built a robust feature-extraction pipeline that improved model context quality and reduced noise from large scan outputs.
• ▸Performed systematic evaluation of LLM approaches (fine-tuning, RAG, chain-of-thought prompting) and identified the most effective configuration for security analysis.
• ▸Designed and deployed a scalable, queue-based processing system that enabled reliable bulk report generation despite limited server capacity.

Focus on machine learning, deep learning, and natural language processing.

As a Software Engineering student at The ICT University in Yaounde, I completed a rigorous program that strengthened my foundations in computer science and practical software development. I worked extensively with algorithms, data structures, databases, operating systems and software architecture while also building real applications across web, mobile and cloud environments. The program trained me to approach problems with an engineering mindset, focusing on scalability, maintainability and clean system design. I also gained experience working on team-based projects, applying industry tools and methodologies to plan, build and deploy software solutions.