About Emmanuel Effiom Duke (Emmanuel Duke)

Background and education

Emmanuel earned his Bachelor of Engineering in Mechanical Engineering from the University of Nigeria, Nsukka in 2024, with a grounding in thermodynamics, energy systems and applied engineering that still shapes his research. Earlier, he received the Agbami Medical & Engineering Scholarship (NNPC/Chevron Joint Venture, 2018). In 2021 he wrote an essay for the Nigeria Higher Education Foundation (NHEF) Essay Competition.

From January 2022 to January 2023 he trained as a student maintenance intern at Transcorp Hotels Plc in Calabar, working on the operation and maintenance of MEP and HVAC systems, including air-cooled chillers, boilers and elevators. In 2024 he was selected for the Politecnico di Torino Earth Summer School, which explored hybrid technologies that convert CO₂ into commercially valuable products. He also completed Imperial College London's Introduction to Air Quality.

AI engineering and AI safety

Emmanuel builds AI products with security at their core. Teger AI, which he created, is an explainable social-engineering defence layer for suspicious email and chat messages. Its Chrome extension scans user-selected Gmail and Slack messages and detects manipulation tactics against a shared threat taxonomy. It then explains the concrete cues behind each warning and recommends a safe next action. Message content is treated as untrusted, the extension never holds provider API keys, and scan history is stored as metadata only. Teger AI is at MVP v0.2 in private beta.

Alongside product work he runs open, reproducible mechanistic-interpretability research:

SPD Mechanism Clustering: a systematic empirical study of automated mechanism clustering for Stochastic Parameter Decomposition. Its seven experiments run from toy models to a real grokked transformer.
Deception Feature Universality: a cross-architecture study that found statistically significant representational similarity for deception cues (CKA 0.56 vs a 0.26 null, p = 0.024).
Reward-Generalization Early Warning: tests whether internal signals can flag unintended reward generalisation before it shows up in outputs.
mechinterp-phishing-probe: probes for social-engineering circuits in language models.

In 2026 his Apart Sprints entry, Four False Positives and One Real Signal: A Controlled Audit of Secret-Loyalty Organisms, placed in the top half of 179 submissions at the Secret Loyalties Hackathon. He has completed AI Safety Hungary's AI Safety Fundamentals course and its Technical AI Safety / Alignment track.

Software engineering and cybersecurity

Emmanuel ships full-stack products across fintech, agriculture, geospatial and security. AxiosPay, a cross-border FX app for African travellers, placed 4th at the Interswitch × Enyata Developer Hackathon. His other public projects include Cropie (crop-disease early warning), SoilIQ (soil-health intelligence), FarmVoice-AI (an agricultural extension agent), GeoDrill (drilling-probability heatmaps) and LandChain Africa (a land registry with GIS boundary-dispute resolution). He holds Cisco's IT Essentials certification. His public code is on GitHub.

Engineering and energy research

His mechanical-engineering research covers computational fluid dynamics, energy optimisation and air quality. Selected work:

The Quadaq Framework: A Multi-Level Partnership Model for Sustainable Air Quality Management, 3rd International Conference on Future Challenges in Sustainable Urban Planning & Territorial Management (SUPTM 2026). doi.org/10.31428/10317/24316.
Experimental Study of Fouling Effect on Feed Water Heaters: A CFD Modeling Approach in Steam Power Plants, with M. N. Eke and I. A. Agunbiade, 10th Virtual International Conference on Science, Technology and Management (2025). doi.org/10.5281/zenodo.14732519.
CFD Simulation of Fouling Effect on Feedwater Heater Outlet Temperature, 2024 Annual European Rheology Conference (2025). doi.org/10.6084/m9.figshare.28280789.v1.
Mass Performance Analysis for Low-power Electric Vehicles in Urban Transportation, with F. O. C. Odunlami, eNergetics 2025. doi.org/10.5281/zenodo.18633459.
Machine Learning-driven Energy Optimization Framework for Hospitality Operations: A Case Study of a 130-Room Hotel, with M. N. Eke, eNergetics 2025. doi.org/10.5281/zenodo.18624282.
Management of the Air For All Open Digital Platform for Community Needs in Citizen Science, Climate and Air Quality Activities, Air For All platform.

He took part in the EU Erasmus+ AirForAll citizen-testing group (2023–2024), and completed Politecnico di Milano's off-grid energy systems modelling course (MICROGRIDSPY) and Strathmore University's Data for Renewable Energy Systems course (both 2025). See the full list on Google Scholar.

Entrepreneurship

Emmanuel founded Dukers LTD, a technology company focused on AI and software products. The initiatives associated with his work are Teger AI and Transly.

Mentoring and community

Since March 2025 Emmanuel has been a Flagship Mentor with the Tony Elumelu Foundation, guiding early-stage African founders. Since June 2026 he has led market research and growth for The Upskill Space, a youth digital-skills academy and Jerry Eze Foundation grantee. He is a leadership alumnus of Common Purpose UK and has been a member of Engineering for Change (E4C) since 2020. In 2023 Street Priests Inc. recognised him with an Award of Active Service.

Profiles

Read his articles and technical notes, or go back to duker.me for ventures, projects, publications and credentials.