Experience
Where the work happened
Five tracks running in parallel rather than one after another: the enterprise AI work, the protocol research, the open source, the evaluation practice, and the client engagements that have been going since 2021.
2025 to present
Employment
AI Infrastructure Engineer
SALI AI
Design, architect, develop, deploy, and maintain enterprise AI systems supporting ESG, sustainability, regulatory compliance, and government digital transformation.
The work runs the whole length of a system, from the architecture through the build to the deployment and the years of maintenance after it, and across four platforms that share one AI layer. It also runs outward: product demonstrations, client workshops, and technical planning with the people who will use what gets built.
Works alongside more than sixteen professionals across engineering, research, product management, and executive leadership, turning ideas from all of them into production software.
- Designed and built DocAI, an internal platform that became the organisation's central AI layer across several enterprise products, and is now being expanded into a standalone platform with memory systems, shared AI services, knowledge management, and intelligent automation
- Architected the backend of the NDPS and FRC regulatory platform supporting Nigerian government operations, covering AI integration, workflow automation, and regulatory intelligence
- Built AI powered sustainability platforms for international initiatives involving UNCTAD and UNEP
- Designed and implemented the ESG analysis capabilities that let organisations process sustainability information through intelligent workflows
- Owned architecture design, AI solution engineering, infrastructure design, feature ownership, and long term platform maintenance
- Ran product demonstrations, client workshops, and technical planning with enterprise clients
What came out of it
- Go
- Python
- FastAPI
- AI Systems
- APIs
- PostgreSQL
- MongoDB
- AWS
- Google Cloud Storage
- Cloudinary
- Docker
- Distributed Systems
2024 to present
Open source
Open Source Engineering
Independent
Creator and lead developer of a set of tools published as free and open source, built as a contribution back to the community rather than as products. There is no commercialisation plan for any of them, and that is the point.
Two of them, Hawk-i and SilkLLM, grew into products in their own right. The rest solve a piece of friction that everyone puts up with and nobody fixes, which is the kind of problem open source is actually for.
- Hawk-i, a security intelligence platform for smart contracts: autonomous exploit generation, AI assisted vulnerability discovery, multi chain analysis, bytecode verification, dependency intelligence, upgrade safety analysis, continuous monitoring, and audit reporting
- SilkLLM, a unified gateway to multiple language model providers behind one key, one balance, and one endpoint, with the metering and billing solved rather than deferred
- The Go-Python Toolchain, three Go binaries that take the slow parts of the Python workflow and do them faster, under Apache 2.0
- DockView and DNI, single container tools that make Docker databases and Docker networks visible with no configuration at all
- WinGate, a privacy layer that shares the windows you meant to share rather than your desktop
- OpenSMTP, a mail engine that sends directly from your own machine so nothing passes through an intermediary
- Anvil Local Suite, a browser extension that gives a local Ethereum node the explorer, wallet, and tooling a public chain has
What came out of it
- Go
- Python
- TypeScript
- Solidity
- Docker
- React
2024 to present
Contract
AI Evaluation and Technical Review
Independent
Evaluating technical output across software engineering and AI systems, which means judging whether something is correct rather than whether it is plausible. The two are easy to confuse and the difference is the entire job.
Covers AI generated code, software architecture, security reasoning, prompt quality, backend and distributed systems, infrastructure design, technical correctness, and AI safety considerations.
Suited to AI evaluation, reinforcement learning from human feedback, model assessment, technical review, and expert annotation work.
- Reviews AI generated code and architecture for correctness, not for fluency
- Assesses security reasoning, which is where a confident wrong answer does the most damage
- Evaluates backend, distributed systems, and infrastructure design against how they behave rather than how they read
- Carries the same standard into his own work: Hawk-i judges a generated exploit by whether it runs, and the deepfake work establishes by ablation which modality is actually doing the deciding
What came out of it
- Python
- Go
- TypeScript
- Solidity
2024 to present
Research
Independent Protocol Research and Security Engineering
Independent
Protocol architecture research, blockchain security analysis, smart contract engineering, and infrastructure design across decentralised finance, blockchain infrastructure, and intent based systems.
Works directly with founders and research teams to turn protocol concepts into secure production systems, which means the research has to end in something that runs rather than in a document.
Research areas span exchange architecture, liquidity pool design, order book systems, market structure, protocol economics, anti MEV research, transaction privacy, zero knowledge research, and smart contract security.
- AngelSwap Protocol, 2024 to 2025: a research focused decentralised exchange ecosystem on PulseChain, covering protocol architecture, production smart contracts, liquidity infrastructure, and order book functionality
- Researched zero knowledge privacy models and protocol level transaction protection, then built them: three circuits, their verifiers, and the contracts that route a shielded trade into public liquidity
- Performed protocol threat analysis and designed the security controls behind configurable privacy mechanisms
- Resolver, 2025 to 2026: protocol architecture, backend engineering, security validation, trust analysis, and attack surface assessment, ending in a working mempool defence rather than a paper
- Go
- Solidity
- Noir
- JavaScript
- Python
2020 to present
Freelance
Freelance Software Engineer
Independent
It started in 2020, writing school projects for coursemates. Django and JavaScript, a real deadline, and someone else's mark riding on it. That is where the habit of finishing came from, and the work has not changed shape since: somebody describes a problem badly, and the job is to turn it into something that works and keeps working after handover.
Since then, backend systems, AI applications, blockchain solutions, smart contracts, APIs, automation, technical consulting, and architecture design for local and international clients.
The range is deliberate. A client with a maize crop, a client with a doctoral thesis on access control, and a client with a mempool problem need different things, and the common part is the same one it was in 2020.
- Began in 2020 building school projects for coursemates in Django and JavaScript, which is where the freelance practice came from
- Amaizing, maize leaf disease identification from a photograph, deployed and running with no paid inference service in the request path
- Multimodal deepfake detection across video, audio, emotion, and metadata, trained on a public benchmark dataset on a single consumer graphics card
- OhiaVault, biometric access control with zero knowledge verification, distributed storage, and on chain anchoring, built for a doctoral research programme
- Resolver, real time sandwich attack detection with rerouting through a private relay
- Backend systems, APIs, automation, technical consulting, and architecture design across local and international engagements
What came out of it
- Python
- Django
- JavaScript
- Go
- Solidity
- TypeScript
- PyTorch
- PostgreSQL