Godwin Olagunju

The Challenge: Navigating Nigeria's tax policies is notoriously difficult for entrepreneurs. The complexity of the regulations often leads to unintentional non-compliance, resulting in severe fines or business shutdowns. There was a need for an accessible tool that could translate legal jargon into actionable business advice. The Solution: I built Lumora, a specialized AI web agent. Unlike generic chatbots, Lumora is built on a RAG (Retrieval-Augmented Generation) architecture. I utilized LangChain to orchestrate the application logic and Pinecone to index vast amounts of official tax data from the CAC. When a user asks a question, the system retrieves the specific legal clauses relevant to the query and passes them to the OpenAI model to generate an accurate, citation-backed answer. This bridges the gap between complex policy and business application.

Engineered a scalable microservices architecture on AWS EKS, utilizing Terraform for Infrastructure as Code to automate full-stack provisioning and cluster management. Designed a secure network topology within AWS VPC by implementing public and private subnets alongside Elastic Load Balancing to optimize traffic distribution for High Availability. Further secured the infrastructure by integrating Amazon RDS for managed persistence, establishing strict access control policies to isolate and protect internal data traffic.

The Challenge: EduPeerHub required a robust, accessible platform to help students from under-resourced backgrounds prepare for exams. The infrastructure needed to be cost-effective yet capable of handling fluctuating traffic without downtime. The Solution: I led the end-to-end deployment of the platform on AWS. My primary focus was automation and performance. I established a secure CI/CD pipeline to bridge the frontend and backend, ensuring that updates could be pushed instantly without disrupting users. To handle traffic spikes, I configured Nginx as a high-performance proxy, which optimized resource usage and request handling. I also secured the platform with end-to-end SSL encryption to protect user data.

The project demonstrates how to build a production-ready CI/CD pipeline that automatically deploys a static website to a cloud Virtual Machine running Ubuntu and NGINX. By leveraging GitHub Actions, the workflow replaces manual file transfers with a secure automation sequence: whenever code is pushed to the repository, the system authenticates via SSH keys, uploads the latest assets to the server, and safely reloads NGINX to reflect changes instantly. This setup provides a "bare metal" alternative to managed platforms like Vercel, giving you full control over your infrastructure while ensuring that your live site is always in sync with your code.

Designed and implemented a blue-green deployment architecture to support zero-downtime releases and rapid rollback. Traffic is routed through Nginx, which automatically fails over between identical production environments based on upstream failures and recovery. The system emphasizes availability over correctness, allowing releases and failures to be isolated without impacting users. Observability is built in through structured access logs, release identifiers, and automated alerts that detect failover, recovery, and elevated error rates. This project demonstrates practical experience in high-availability system design, traffic routing and failover mechanics, and operational visibility under failure conditions.

Played a pivotal role in the engineering team for Sitelytics by architecting the server-side foundation of the application. While partnering with mobile developers to define data models and API specifications, focused heavily on DevOps enablement—automating deployment workflows, provisioning cloud resources via Ansible, and implementing centralized logging to give the team visibility into system performance and AI scanning processes.

Developed a production-ready bash script that automates end-to-end deployment of Dockerized applications to remote Linux servers, emphasizing reliability, user experience, and error recovery. The solution implements a cohesive defense system through input validation with confirmation loops, comprehensive timestamped logging, idempotent operations for safe re-execution, and semantic error handling with granular exit codes. The script orchestrates multi-stage deployment workflows including GitHub authentication, SSH-based remote server provisioning, Docker container orchestration, Nginx reverse proxy configuration, and automated SSL certificate management via Certbot. Built with human-centered design principles, the script reduces cognitive load during high-pressure deployments through transparent feedback loops and graceful failure recovery, making it reliable for use in production environments. Technologies: Bash scripting, Docker, Docker Compose, Nginx, Git, SSH, SSL/TLS, Linux system administration.