Backend engineer with five years of professional experience building production systems across FinTech, Healthcare, and E-commerce. My work is weighted toward AI integration — specifically designing systems where LLMs do real clinical or financial work, not demos.
Most recently at ToksNet, I led backend architecture for the Adeon AI Healthcare Platform: a 3-tier microservices system integrating Gemini 2.5 Pro and OpenAI Whisper for real-time medical transcription and automated condition analysis. Before that, Senior Laravel developer at Sosotoba Institute of Technology, leading a team building hospital management software on MySQL databases handling millions of records.
B.Eng in Computer Engineering from the University of Benin. Fullstack certification from ALX Africa.
AI & Machine Learning
LangChain LLMs — Gemini 2.5 Pro · GPT-4 Whisper SHAP LIME RAG Pipelines FAISS
Backend Engineering
FastAPI Node.js Laravel Express REST & GraphQL APIs WebSocket Microservices
Mobile
Flutter Dart React Native Expo
Cloud & DevOps
Docker AWS S3 · Transcribe Google Cloud Platform CI/CD Nginx
Languages
Databases
PostgreSQL MySQL Redis FAISS
|
Adeon AI Healthcare Platform ToksNet · Oct 2025 – Jan 2026 · Private 3-tier microservices platform integrating Gemini 2.5 Pro and Whisper for automated medical transcription and skin/eye condition analysis. 20+ FastAPI endpoints with LangChain managing conversational context across sessions. Achieved 2.5–2.9× real-time transcription speed via multi-level caching and async architecture. Deployed with Docker, circuit breaker patterns, and HIPAA-conscious content safety pipelines.
|
Crime Prediction & Explainable AI Machine learning pipeline for crime pattern prediction built around transparent decision-making. TensorFlow model with SHAP and LIME post-hoc explainability, surfaced via a Streamlit dashboard that lets analysts interrogate individual predictions rather than treat the model as a black box.
|
|
Keystroke Dynamic Digital Twin Statistical digital twin for behavioural biometric authentication. Anomaly detection on keystroke dynamics using Isolation Forest, One-Class SVM, and distance-based baselines — evaluated against the CMU benchmark dataset. University of Lincoln, CMP3753.
|
Lincoln Business Suitability Index Spatial analytics pipeline analysing 13,097 businesses across 55 H3 hexagonal zones in Lincoln, UK. Ingests ONS Census 2021 and OpenStreetMap data, computes a composite Business Suitability Index, and surfaces results through a Flask API and React map dashboard.
|
|
GiftyView — Gift Card & Payments Platform Active Production · App · API Production FinTech platform for global gift card trading, crypto exchange, and utility bill payments. Flutter cross-platform app backed by a TypeScript/Prisma API integrating Flutterwave, Paystack, and Korapay for fiat payments; Bitnob, Quidax, and NowPayments for crypto; Reloadly and Prestmit for gift card fulfilment; VTPass for bills; SmileID for KYC identity verification; OpenAI for AI-powered support; and Redis for caching. Three separate deployments — consumer app, backend API, and admin portal.
|
Lendsqr Digital Wallet Backend Production-grade REST API for a mobile lending platform: account creation, wallet funding, peer-to-peer transfers, and withdrawals with real-time blacklist screening via the Lendsqr Adjutor API. Node.js, TypeScript, KnexJS on MySQL with full unit test coverage.
|
|
Algorithmic Trading Suite BA-bot-MT5 · Gold Breakout Bot · Crypto Signal Bot · Telegram Bot Four live trading systems across two stacks. BA-bot-MT5 is a production MQL5 Expert Advisor with modular architecture — BrokerDetector, NewsFilter, RiskManager, Signals engine, Statistics, and charting — achieving a 65% signal accuracy in live conditions. The Python suite covers gold breakout detection, SMC-based crypto signal generation, and Telegram-native real-time alerting.
|
Football Performance Predictor Machine learning pipeline predicting football player performance from historical FPL statistics. End-to-end system: data ingestion, feature engineering, model training, and an interactive web interface for querying predictions. Built to demonstrate applied ML on real-world sports analytics data.
|
|
Iron Discipline (In Development) Personal fitness transformation OS built in React Native and Expo. Offline-first architecture with SQLite persistence and Zustand state management. Features a custom daily discipline scoring engine, progressive overload tracking, macro and supplement logging, habit checklists with haptic feedback, and progress photo analytics with trend charts. Designed as a single system to replace every fragmented fitness tool.
|
| Period | Role | Organisation |
|---|---|---|
| Oct 2025 – Jan 2026 | AI Engineer · Lead Backend Developer | ToksNet (Remote) |
| Aug 2023 – Feb 2024 | Senior Laravel Developer | Sosotoba Institute of Technology |
| Jun 2022 – Jul 2023 | Junior Software Developer | Tryve Technologies |
B.Eng Computer Engineering — University of Benin, Nigeria · 2024
Fullstack Web Developer — ALX Africa · 2023
Email · [email protected]
Website · eghosa.netlify.app
WhatsApp · Direct message
