|

“The scientist describes what is; the engineer creates what never was.”

— Theodore von Kármán

// ABOUT ME

Builder. Tinkerer.
Still figuring things out.

I’m Manasseh — a machine learning engineer, tinkerer, and founder of MVM Labs. I build systems that learn, adapt, and hopefully do something useful outside a notebook.

I spend a lot of time experimenting with ideas, chasing curiosities, and building things just to see where they lead. Sometimes that means training models, sometimes writing software, sometimes sketching out strange systems that sit somewhere between engineering, robotics, and research.

This site is basically my workshop bench on the internet. A place for projects, notes, experiments, half-formed ideas, the things I keep tinkering with as I grow into the kind of builder I want to become, and the occasional rabbit hole I probably didn’t plan to go down.

Field note · 01

Systems grow through iteration.
Curiosity shapes the path.

§02

EXPERIENCE

Founder / Builder

MVM Labs

2026 -- Present
Nairobi, Kenya

Mind to understand. Vision to imagine. Motion to build.

SoftwareMachine LearningResearchEngineering

Software Engineer

Digipedia Limited . Contract

Jun 2025 -- Mar 2026
Nairobi County, Kenya . Hybrid

Working on software infrastructure and ERP software: backend systems, business workflows, integrations, internal tools, and scalable software foundations.

ERP SoftwareBackend SystemsInfrastructureAPIsDatabases

Software Developer

Strathmore Business School (SBS) . Contract

Jan 2024 -- Jul 2025
Nairobi County, Kenya . On-site

Built and maintained software systems for institutional operations, focusing on system development, automation, internal tools, and practical software delivery.

System DevelopmentFull-stackInternal ToolsAutomation
SKILLS & TOOLSHIGHLIGHTS ACTIVE ROLE
PROGRAMMING LANGUAGES
PythonPHPTypeScriptGoJavaScript
FRAMEWORKS & LIBRARIES
FastAPILaravelFlaskNext.jsTensorFlowOpenCV
BACKEND & DATA SYSTEMS
REST API DesignSystem IntegrationPostgreSQLRedisVector Databases
AI & APPLIED MACHINE LEARNING
Retrieval-Augmented Generation (RAG)Representation LearningCNNsVision TransformersReinforcement LearningIntelligent Agent Systems
INFRASTRUCTURE, DEPLOYMENT & TOOLING
DockerAWS EC2AWS S3AWS LambdaSemantic SearchHybrid Retrieval SystemsLLM IntegrationOllamaLLaMA
§03

PROJECTS

A selection of systems I’ve built or explored — from developer tools and AI orchestration to marketplace software and intelligent environment understanding.

PROJECT
01/ 04
CATEGORYCOMPUTER SCIENCE / DEV TOOLING
STATUS ACTIVE
UPDATEDJUN 2026
CURRENT PROJECT

Big O Lab

Algorithm analysis playground

Big O Lab is an interactive algorithm analysis playground that lets users write Python code, run experiments across different input sizes, and visualize runtime complexity directly from the browser.

A browser-based workspace for learning and exploring algorithmic performance. Users can write or select algorithms, configure input profiles, run empirical experiments, inspect line-level runtime behavior, compare experiments, and generate complexity explanations using heuristic analysis or optional LLM support.

View Project ↗Read Case Study ➔
SYSTEM EMPHASIS / TECHNICAL NOTE

Combines algorithms, execution instrumentation, backend systems, charts, sandboxing, rate limits, database persistence, and optional AI explanations.

STACK

Next.js · Monaco Editor · Recharts · Zustand · FastAPI · PostgreSQL · Redis · Ollama Cloud

AlgorithmsDeveloper ToolsFastAPINext.jsRuntime AnalysisLLM Explanations
SYSTEM DIAGRAMFIG. 03.01
PLAYGROUNDMonaco EditorRecharts / UIZustand StoreFRONTENDFASTAPI APIPydantic / CORSRate Limits / AuthROUTINGEXEC RUNNERPython SandboxTime InstrumentsINSTRUMENTCOMPLEXITY ENGEmpirical HeuristicsBig O AnalysisESTIMATORPERSIST / AIPostgreSQL (DB)Redis CacheOllama Cloud (LLM)EXPERIMENTAL INSTRUMENTATION WORKFLOWInput generation runs code dynamically, estimates runtime complexity curve,then renders Recharts graph & calls LLM Cloud for heuristics review.
§04

RESEARCH

RESEARCH INDEX
PAPER 001GLOBAL UNDERGRADUATE AWARDS · HIGHLY COMMENDED · 2025
STATUS
Highly Commended➔

Early Detection of Diabetic Retinopathy Using EfficientNet-Based Convolutional Neural Networks

AREAMedical AI · Computer Vision · Deep Learning
INSTITUTIONStrathmore University
REGIONAfrica & Middle East
AUTHORManasseh Maina
DATE / YEAR2025
ABSTRACT

Diabetic retinopathy is a diabetes-related eye disease and a major cause of preventable blindness, especially in regions with limited access to ophthalmologists. This research developed an EfficientNet-B0-based convolutional neural network to detect and classify diabetic retinopathy into five severity levels: No DR, Mild, Moderate, Severe, and Proliferative. The system used a balanced dataset of 10,000 retinal images, with preprocessing techniques including CLAHE, resizing, and data augmentation to improve image quality and model robustness. Transfer learning and attention-based feature extraction were used to improve diagnostic sensitivity. The model achieved 94.97% accuracy, 94.84% sensitivity, and 95.1% specificity. It was integrated into a Flask-based web application with Grad-CAM visualization, providing interpretable model outputs for physicians and supporting scalable, cost-effective screening in resource-limited healthcare settings.

KEYWORDS
EfficientNet-B0Diabetic RetinopathyMedical ImagingCNNsGrad-CAMFlaskCLAHETransfer Learning
KEY RESULTS
Accuracy94.97%
Sensitivity94.84%
Specificity95.1%
Dataset10,000 images
Classes5 DR Levels
RESEARCH QUESTIONS
→How can deep learning improve early detection of diabetic retinopathy from retinal images?
→Can EfficientNet-B0 classify diabetic retinopathy severity levels with high accuracy, sensitivity, and specificity?
→How can model interpretability tools like Grad-CAM improve physician trust in AI-assisted diagnosis?
→How can medical AI systems be deployed practically in resource-limited healthcare environments?
KEY CONTRIBUTION

An interpretable medical AI screening system that combines EfficientNet-based deep learning, retinal image preprocessing, five-class disease severity classification, and Grad-CAM visualization to support early diabetic retinopathy detection in resource-limited settings.

FIG. R-01 — DIAGNOSTIC PIPELINE
Retinal ImagePreprocessingCLAHE · ResizeEfficientNet-B0Transfer LearningClassification5 DR Severity LevelsGrad-CAM MapInterpretabilityPhysician ReviewClinical Decision
READ PAPER RESEARCH NOTES
§05

CONTACT

Let's build something
worth building.

I'm open to conversations about AI, robotics, emergent software, aerospace software, research, and building serious technology from Africa. If it is technical, ambitious, and worth doing, reach out.

+
+
SYS_REF // COMM
DIRECTmanassehmabuya@gmail.com
CLICK TO COMPOSE
SYS_REF // GEO// -1.2865, 36.8208
LOCATIONNairobi, Kenya
GMT +3 (EAT)
SYS_REF // STAT[OPEN]
OPEN TO
＋Collaborations
＋Research conversations
＋Technical projects
AVAILABILITY 2026

— End of field notes.