Dukers LTD
The technology company I founded, and the home of my product work in AI and software. Business enquiries go to emmanuel.duke@dukersltd.com.
I build AI products, engineer systems and do research. I founded Dukers LTD, and my work spans AI safety and cybersecurity, full-stack software, and energy and air-quality research.
AI Engineer · Mechanical Engineer · Researcher · Founder of Dukers LTD
Ventures
The technology company I founded, and the home of my product work in AI and software. Business enquiries go to emmanuel.duke@dukersltd.com.
An explainable social-engineering defence layer for suspicious email and chat messages. A Chrome extension scans Gmail and Slack messages, flags manipulation tactics with the cues behind each warning, and recommends a safe next action. MVP v0.2, private beta.
A product initiative I'm associated with. See transly.software for current details and availability.
Selected work
Cross-border FX for African travelers — swap NGN, KES, UGX, GHS & ZAR at live mid-market rates with a 1.5% flat fee and instant settlement. 4th place, Interswitch × Enyata Developer Hackathon.
Explainable social-engineering and phishing detection for Gmail and Slack: a Chrome extension and API that surface manipulation tactics, the cues behind them and a safe next action. MVP v0.2, private beta.
Mechanistic-interpretability research: the first systematic empirical study of automated mechanism clustering for Stochastic Parameter Decomposition (SPD). Seven experiments from toy models to a real grokked transformer, showing component specialization — not clustering-signal choice — is the binding constraint. Paper + reproducible code.
Reproduces unintended reward generalization (the OpenAI 'goblin' dynamic) in a controlled model and tests whether internal signals can flag reward spread before it surfaces in outputs — a follow-up to OpenAI's alignment research.
Cross-architecture interpretability study finding statistically significant representational universality (CKA 0.56 vs null 0.26, p=0.024) for deception cues — evidence that different models represent phishing/social-engineering concepts similarly.
AI-powered geological drilling intelligence platform — scores drilling probability across a region and renders a live risk heatmap.
Arduino-based motion-sensing alarm using a PIR sensor and buzzer to detect movement and trigger an alert.
Mechanistic interpretability to detect social-engineering circuits in LLMs.
More on GitHub.
About
I'm Emmanuel Effiom Duke (also Emmanuel Duke), an AI engineer, mechanical engineer and technology entrepreneur. I hold a B.Eng. in Mechanical Engineering from the University of Nigeria, Nsukka, and I'm the founder of Dukers LTD.
AI engineering and safety. I build AI products such as Teger AI, which targets psychological manipulation in messages. I also run mechanistic-interpretability research on how models represent deception and reward, published as open, reproducible code.
Software and cybersecurity. I ship full-stack products in fintech, agriculture, geospatial and security. They include AxiosPay, a cross-border FX app that placed 4th at the Interswitch × Enyata Developer Hackathon.
Engineering research. My mechanical-engineering work covers CFD modelling of steam-plant feedwater heaters, machine-learning energy optimisation and air-quality management, presented at international conferences.
Community. I mentor African founders through the Tony Elumelu Foundation and lead growth for The Upskill Space, a youth digital-skills academy.
Read the full biography of Emmanuel Effiom Duke →
Experience
The Upskill Space [Your Digital Skill Academy] · Part-time · Enugu State, Nigeria
Jun 2026 – Present
Drives growth strategy for a fast-scaling upskilling academy — turning student behavior, competitor moves, and market trends into decisions that grow reach and retention.
The Tony Elumelu Foundation · Part-time
Mar 2025 – Present
Mentors early-stage founders across Africa through the continent's flagship entrepreneurship programme, helping them sharpen ideas into fundable, resilient businesses.
Engineering for Change (E4C) · Part-time · Remote
Apr 2020 – Present
An active member of a global engineering community applying technology to real-world development challenges in underserved regions.
EU Erasmus+ AirforAll · Seasonal
Feb 2023 – Feb 2024
Joined an international citizen-testing group to grow public air-quality knowledge and stress-test the usability of an open environmental toolkit under the EU Erasmus+ AirForAll project.
Transcorp Hotels Plc · Internship · Calabar, Nigeria
Jan 2022 – Jan 2023
Hands-on training in the operation and maintenance of MEP and HVAC systems — air-cooled chillers, boilers, and elevators — in a live hospitality environment.
Education
B.Eng, Mechanical Engineering — foundation in thermodynamics, energy systems, and applied engineering that underpins my research.
Selected for the Earth Summer School, exploring hybrid technologies that convert CO₂ into commercially valuable products (2024).
Examined the health impacts of urban air pollution and the science-led strategies used to measure and reduce it.
Speaking & Community
Head of Market Research & Growth (Volunteer) · Jerry Eze Foundation Grantee
Volunteer growth lead for a youth-focused upskilling enterprise backed by the Jerry Eze Foundation — shaping the strategy that widens access to digital skills.
Mentor
One of the mentors guiding African entrepreneurs through the Tony Elumelu Foundation — the continent's largest philanthropic entrepreneurship platform.
Leadership Alumnus
Graduate of Common Purpose UK's leadership programme, trained to lead across cultures, sectors, and boundaries.
How I can help
Whether you're a research group, a startup, a climate/energy programme, or a mission-driven organization, here's where I can add value — grounded in real research and shipped work.
CFD modeling, energy-systems analysis, ML-driven optimization, and citizen-science air-quality platforms — from study design to publication.
Ship real products end to end — fintech, AI agents, geospatial dashboards, and data platforms across web and backend.
Turn user behavior, competitor analysis, and market trends into growth decisions that expand reach and retention.
Data collection, analysis, and machine-learning modeling for energy, climate, and operational efficiency problems.
Mentor founders and students (Tony Elumelu Foundation, The Upskill Space) and lead community upskilling programmes.
Common Purpose-trained to work across cultures, sectors, and disciplines — bridging research, industry, and community.
Explore by area: AI engineer · AI safety researcher · AI security engineer · AI forecasting and LLM evaluation · Mechanical engineer and energy researcher · all expertise
Research
Mechanism clustering for Stochastic Parameter Decomposition, cross-architecture universality of deception features, and early warning for unintended reward generalisation. Code is public on GitHub.
CFD modelling of fouling in steam-plant feedwater heaters, and performance analysis of low-power electric vehicles.
A machine-learning energy-optimisation framework for hospitality operations, built from a 130-room hotel case study.
Multi-level partnership models for air-quality management and open digital platforms for citizen science (EU Erasmus+ AirForAll).
Full list on Google Scholar and ORCID. This list syncs from ORCID weekly.
Writing
Forecasting · · 3 min read
Lessons from an LLM forecasting bot in Metaculus AI tournaments: silent research failures, extremized toss-ups, and replacing vibes-tuning with measurement.
AI evaluation · · 2 min read
Measuring nine language models' real knowledge cutoffs with dated-event bisection: every model that self-reported a cutoff was wrong, some by over a year.
AI security · · 2 min read
How Teger AI, an explainable social-engineering defence for Gmail and Slack, pairs deterministic signals with AI analysis and treats every message as untrusted.
All articles at duker.me/articles.
Recognition
Apart Sprints — Four False Positives and One Real Signal: A Controlled Audit of Secret-Loyalty Organisms placed in the top half of 179 submissions.
Street Priests Inc. — certificate of recognition.
NNPC/Chevron Joint Venture.
State-level winner, Science Competition for Primary Schools.
Participant (NTS).
MICROGRIDSPY — energy systems modelling (2025).
Data analysis & collection for renewables (2025).
Introductory AI Safety Fundamentals course and the Technical AI Safety / Alignment track.
Leadership, relationship building & collaboration (2020).
UNITeS Cisco Networking Academy.
Business and operations foundations.
Full profile on LinkedIn.