Nourish Neural
Does a heterogeneous ML ensemble plus a local LLM agent make a more useful food waste tool than either alone?
- Python
- TypeScript
- React
- FastAPI
- PyTorch
- scikit-learn
- Ollama
- Llama 3.2 via Ollama
- RAG
- Tool use and function calling
AI and software engineer
I build AI systems for settings where mistakes have consequences, from clinical portals to local LLM agents.
10+
freelance clients
E-commerce, professional services and retail
98%
ensemble accuracy
On 10,000 synthetic samples, not real-world data
100%
test pass rate
Python and Node.js end-to-end tests on healthcare portals
1st
place, solo
RUNCHOW Hackathon, built in 6 hours
100+
society members led
AI Innovators Society, UWE Bristol
Projects
Three flagship projects with case studies, then the rest of what I am building. Each case study lists its limits.
Does a heterogeneous ML ensemble plus a local LLM agent make a more useful food waste tool than either alone?
Voice-enabled meal logging with AI scoring, built solo in 6 hours.
An LLM that queries live CVE data, so vulnerability checks happen in context.
PlantVillage EDA on 54K images across 38 classes, a fine-tuned ResNet-50 with Grad-CAM, a LangGraph agronomist agent and an XGBoost yield predictor with SHAP.
An LSTM autoencoder for unsupervised anomaly detection on multivariate NASA and ESA telemetry, with an Isolation Forest baseline and a plain-English query interface that produces Plotly charts.
An AI triage and routing agent. An XGBoost triage classifier with SHAP feeds a LangGraph agent that checks NICE guidelines, asks clarifying questions below 70% confidence, and keeps an audit trail for every routing decision.
Designed parallel race routing between on-device and cloud models at the Google DeepMind x Cactus Hackathon at UCL East, February 2026. I took part, I did not win.
IoT system in C++ with UDP communication and Catch2 tests.
An AI assistant built in TypeScript.
An AI medical assistant built with Python and NLP.
A Python simulator for AI drone design.
Timeline
Production work in healthcare and freelance AI, alongside leadership roles at UWE Bristol.
May 2026 to present
Public Health Bureau
Building sovereign, NDPR-compliant clinical AI systems for clinicians across Nigeria.
Aug 2025 to present
Fiverr
10+ clients across e-commerce, professional services and retail, averaging a 40% productivity improvement per client.
2025 to 2026
AI Innovators Society, UWE Bristol
Led an 8-person committee and 100+ student members through hackathons, workshops and speaker events.
April 2026
Tech Nexus / Maendeleo Africa Fellowship Institute
Designed and delivered a one-month programme on tech fundamentals, data literacy and cybersecurity.
Sep 2025 to May 2026
UWE Bristol Careers Service
Coached students on career planning across software engineering, data and AI roles.
Sep 2024 to May 2025
UWE Bristol
Weekly Python and data structures and algorithms sessions for first-year students.
Oct 2023 to May 2024
UWE Bristol
Represented 200+ CS students and turned their feedback into curriculum and support policy proposals.
Toolkit
Select a skill to see where I have used it.
Recognition
Hackathon results, society leadership and student roles.
Award
Built FuelFitAI solo in 6 hours, beating full teams.
Recognition
Dependency Watcher, November 2025.
Participation
UCL East, February 2026. Designed parallel race routing between on-device and cloud models.
Leadership
Led an 8-person committee and ran hackathons, workshops and speaker events for 100+ members.
Leadership
Weekly Python and data structures and algorithms sessions. Engagement up 30%.
Leadership
Represented 200+ students and turned feedback into curriculum and support policy proposals.
Background
I build AI for settings where mistakes have consequences. At the Public Health Bureau I build clinical AI portals with human review built in, local retrieval for offline-resilient access to medical guidelines, and a de-identification layer that strips patient details before anything leaves the device.
Before that I shipped AI tools for freelance clients. A sentiment engine cut complaint response time by 60%. An assistant automated 80% of customer enquiries. My dissertation asked whether an ML ensemble and a local LLM agent beat either one alone.
I am working toward an MSc in Information Technology at UWE Bristol. I am looking for graduate roles and internships in software engineering, AI and ML engineering, and data science, ideally with teams working on problems that matter.
Study
Sep 2026 to present
UWE Bristol
Sep 2023 to Jul 2026
UWE Bristol
First Class Honours
Get in touch
Graduate roles, internships and collaborations on AI and software systems.
Recently updated public repositories, cached for an hour.
Share of repositories, not lines of code.