AI Robotics Engineer
Independent Robotics & AI Systems Engineer specialising in autonomous systems, computer vision, and AI-driven robotics. Turning research into real-world deployments.
Tech Stack
Specialisation
Status
Open to work
Featured Project
I'm Iyanuoluwa Enoch Oke — an independent Robotics & AI Systems Engineer building systems at the intersection of robotics, computer vision, and artificial intelligence.
My work spans the full stack of intelligent systems: from low-level motor control and sensor fusion in ROS2 environments to high-level AI decision layers using modern deep learning architectures and multi-agent frameworks.
I don't just build demos — I ship production-grade systems. Sentinel Quant is a live-deployed AI algorithmic trading system managing real capital across two brokers, with full system monitoring, automated audit pipelines, and a private React dashboard. Signlytic AI is a bidirectional BSL translation system engineered end-to-end. Every project is built to production standard.
Areas of Focus
A full-stack engineering skill set spanning robotics hardware, intelligent perception, and AI systems.
A focused selection of production systems and engineering experiments.
Bidirectional British Sign Language Translation System
Real-time BSL translation system using MediaPipe Holistic and Video-SWIN-T trained on 5,200+ sign sequences — achieving 100% Top-1 accuracy. Deployed as a Chrome extension with a Three.js 3D avatar and four caption integrations.
Live AI Algorithmic Trading System
Dual-broker AI trading system managing live capital across Alpaca and Trading 212 with Docker orchestration, VIX regime detection, three autonomous strategy sleeves, and a 138-check automated audit.
CARLA Simulation
ROS 2 perception and localisation stack integrating YOLOv11, ByteTrack, StrongSORT, and visual SLAM in CARLA — sustaining 25–35 FPS with ~0.9m median localisation error in urban driving scenes.
ZetaBot
Camera-assisted 6-DOF robotic arm built with Arduino Nano, ESP32, and a ROS serial bridge on a solar-supported power architecture — achieving sub-2° RMS joint tracking error during tested motion profiles.
Meta-Agent Workflow Compiler
Meta-agent framework that compiles YAML-defined specifications into canonical JSON workflows, Mermaid diagrams, and runnable Python agents — with deterministic, auditable validation and code generation pipelines.
Multi-Agent Code Refactoring System
AI-powered swarm system for autonomous codebase analysis, refactoring, testing, and documentation generation using multi-agent orchestration.
Peer-archived research in AI governance, robotic perception, and embedded robotic systems.
Oke, I. E. · Zenodo
Oke, I. E. · Zenodo
Oke Iyanuoluwa Enoch · Zenodo
Oke Iyanuoluwa Enoch · Zenodo
Open source contribution, academic peer review, and technical writing published by leading engineering platforms.
MervinPraison/PraisonAI
Among the top 20 contributors to PraisonAI, an open-source multi-agent AI framework with over 8,000 GitHub stars. My contributions included improvements to agent termination and telemetry correctness, Windows compatibility, LiteLLM fallback support, and substantial hardening of the integration test suite.
Scientific Reports · BMC Artificial Intelligence
Awarded two Springer Nature reviewer certificates in 2026 for peer review contributions to Scientific Reports (Nature Portfolio) and BMC Artificial Intelligence, assessing submitted manuscripts for scientific rigour, methodology, and contribution to the field.
UPES Dehradun · Procedia Computer Science (Elsevier)
Served as external reviewer and assessor for the 4th International Conference on Machine Learning and Data Engineering (ICMLDE 2025), held 6-8 November 2025 at the School of Computer Science, UPES Dehradun, India, with proceedings published in Elsevier's Procedia Computer Science. Recognised by the Organising Chair, the Dean of the School of Computer Science, and the Vice Chancellor of UPES.
A full tutorial on building a production-ready robotic perception system: ROS 2 integration, YOLOv11 detection, ByteTrack multi-object tracking, and ONNX optimisation for deployment on edge hardware. Published by freeCodeCamp and promoted to their 2.2M+ followers on X and LinkedIn.
The design principle behind every AI system I ship: AI proposes, deterministic rules validate, humans intervene where necessary, and everything important is logged. Covers why a model can be highly accurate and still unsafe as an authority, drawn from applying the pattern across robotics, computer vision, language systems, and quantitative finance.
The story behind Signlytic, a Chrome extension that turns live captions into British Sign Language animations in real time, and the case for building accessibility tooling without waiting for permission.
Practical lessons from integrating YOLOv11, ByteTrack, StrongSORT, CARLA, and ROS 2 into a real-time perception system, and why the hard problems are systems-level rather than model-level.
Why the gap between model performance and system reliability is where most AI products quietly collapse. Argues that models are probabilistic components rather than decision-makers, drawing on the human-in-the-loop architecture behind Sentinel Quant and Signlytic.
What clean benchmarks and polished simulator runs conceal: tracker drift, creeping latency, and edge devices throttling under real load. Lessons from taking ROS 2, YOLOv11, SLAM Toolbox, ByteTrack, and CARLA out of simulation and onto real hardware with ONNX and TensorRT.
Iyanuoluwa Enoch Oke
Robotics & AI Engineer
Open to full-time roles, research collaborations, and freelance projects in robotics, AI systems engineering, and computer vision. Based in the UK, available globally.
Send me a messageIyanuoluwa Oke
© 2026 — Independent Robotics & AI Systems Engineer
