20+ AI, software and infrastructure projects delivered since 2019 — spanning fintech, healthcare, marketplaces, education and security, for clients in Ghana, USA, Canada and Equatorial Guinea.
AI interview-intelligence and career-readiness platform (Interview Prep Agent / Job Search Assistant). Built with FastAPI, React, TypeScript, MongoDB and RAG pipelines powered by multi-model orchestration. Designed context-aware retrieval and evaluation workflows to provide real-time candidate feedback during technical interview preparation.
Conversational e-commerce search engine (RAG Shopping Assistant) using a RAG architecture and vector retrieval grounded strictly in live database inventory. Eliminates hallucinated recommendations and solves natural language query misunderstanding in traditional keyword-based search.
Diaspora marketplace built for a Canadian client — designed service schemas, backend data models, and API integration patterns for merchants, inventory, orders, payments, delivery and fulfilment for a fully remote team.
Food delivery (AhotoEats) and healthcare/medicine-delivery (AhotoCare) apps for the Ghanaian market, released on Play Store and App Store across four iOS and four Android builds.
Data-mining and AI-assisted lead-research platform built with FastAPI, SQL, and LLM orchestration — handles data extraction, lineage tracking, automated retries, and quarantine reprocessing for research teams.
Singly developed the Node.js backend, API documentation, integration, VPS server setup/configuration and deployment for an affordable data-bundle commerce platform. Part of a Node.js/MongoDB commerce operation handling USD 500,000+ at 99.7% uptime.
Backend development, API integration and system administration for a trading data platform — real-time transaction dashboard, pricing and bulk data-share tooling.
Node.js services, access controls, transaction workflows and VPS deployment for an airtime-and-data distribution platform (SawTel) that processed 100,000+ secure transactions with no recorded security incident.
Distributed Python services, model APIs and financial-data pipelines for a commodity-trading platform — LSTM and Hidden Markov Model workflows from data prep through production monitoring.
Conversational assistant using NLP, RAG and LLM integration, grounding responses in connected knowledge sources.
Responsive website and custom CMS for EBEYE General Services — achieved 100% direct publishing autonomy for non-technical staff and reduced content update turnaround from days to minutes.
Trained by the Ghana Statistical Service (GSS) team on acquiring, cleaning, and analyzing data from Statsbank.com. Combined GSS national statistical datasets with real-time IoT environmental sensor feeds using CatBoost ML models — awarded regional second place at the Ghana Statistical Service Data Hackathon 2024.
ESP32 and NodeMCU access-control solutions with C++ firmware, encrypted transmission and PHP/MySQL administration — deployed live to more than one organisation.
Predictive analytics evaluating startup funding potential from historical investment and business data.
Application logic, REST APIs and integration layers for a microservices-based campus-navigation platform; optimised database access and wrote API documentation for handover.
Responsive website and custom CMS for Boten Limited, letting non-technical editorial staff publish and maintain content directly without developer intervention.
An online used and new car dealership platform — vehicle inventory browsing, financing pre-approval, test-drive scheduling and a guided purchase flow.
Desktop application for records, grading, fees, parent communication and role-based administration.
Researched query routing and semantic translation techniques to transform natural-language inputs directly into optimized downstream database queries. Addressed token inefficiency and latency bottlenecks in standard LLM-to-database interfaces. Published research findings on ResearchGate, contributing novel frameworks for low-latency LLM query execution.
Designed multi-turn stress-testing prompts and automated evaluation models to measure LLM truth retention and hallucination resistance under adversarial stress. Investigated structural causes of conversational drift in Large Language Models and documented a repeatable red-teaming framework for benchmarking and mitigating conversational hallucinations.
Designed a modular protocol for distributed multi-agent ecosystems covering capability discovery, gRPC task delegation, RAG memory, and vector search. Built and published the official CoNET Python package on PyPI as open-source agent infrastructure.
Built a venture assessment system combining NLP narrative extraction, RAG pipelines, and predictive models using Python, Pandas, Scikit-learn, and SQL. Automated data-driven risk evaluation for early-stage startup investments and standardized startup pitch analysis through combined statistical modeling and automated pitch deck evaluation.
Trained directly by the Ghana Statistical Service (GSS) team on acquiring, cleaning, and analyzing data from Statsbank.com. Combined GSS national statistical datasets with real-time IoT environmental sensor feeds using CatBoost ML models for live environmental classification and predictive analysis — regional second place at the Ghana Statistical Service Data Hackathon 2024.
Additional projects covering commerce platforms, business tooling and security utilities.
Open to full-time roles, contract work and relocation.