Godwin Olagunju

As an AI Engineer at Peepalytics, I apply advanced machine learning techniques across natural language processing and computer vision to deliver impactful solutions that drive significant business growth and operational efficiency.

• Designed and implemented an automated data pipeline, streamlining data processing and analysis, which reduced manual efforts by up to 66% (3x) and enabled efficient insight generation for users.
• Developed a crowd density prediction model, providing real-time event attendance estimates that enhanced event management and optimized planning and resource allocation.
• Engineered an agentic chatbot integrating data from multiple sources, facilitating fast and accurate queries to significantly enhance organizational data access and decision-making processes.
• Built a voice agent to reduce missed calls by 50% and increase customer engagement, contributing to higher satisfaction and retention rates.

At Skoutwatch, I specialized in developing and deploying state-of-the-art computer vision models, implementing custom training pipelines, and architecting scalable cloud deployment strategies.

• Developed and fine-tuned advanced computer vision models (SAM2, YOLO) for video analysis, achieving a 35% increase in detection accuracy across challenging environments and lighting conditions.
• Implemented end-to-end custom training pipelines, incorporating data augmentation and hyperparameter optimization, which reduced false positives by 28% while maintaining processing speed.
• Built comprehensive RESTful backend endpoints using FastAPI, serving computer vision models at scale with caching mechanisms and optimized inference paths for high-throughput applications.
• Designed and implemented advanced batching processes for data pipelines, leveraging parallel processing to optimize throughput by 40% and significantly reduce processing latency.
• Architected a cost-efficient cloud deployment strategy using AWS Lambda and EC2 instances, employing event-driven architecture for seamless scaling during peak demand.

As a DevOps Engineer at HNG, I developed CI/CD pipelines, managed server infrastructure, and established robust monitoring ecosystems to ensure high availability and operational efficiency.

• Developed CI/CD pipelines with GitHub Actions, automating testing and deployment to reduce release cycles from days to hours while ensuring code quality.
• Managed server infrastructure across diverse environments, implementing configuration management and Infrastructure-as-Code practices to achieve 99.9% uptime.
• Established a comprehensive monitoring ecosystem using Prometheus, Grafana, and Alert Manager, reducing system incidents by 65% through proactive Slack notifications.
• Configured Nginx as a reverse proxy with caching, load balancing, and SSL termination, improving application response times by 30% and enhancing security.
• Developed automation scripts to standardize environments, significantly reducing configuration drift across development, staging, and production.

Lumora is an intelligent web chat agent designed to assist business owners in navigating Nigeria's complex tax regulations. Many entrepreneurs struggle to understand the policies that govern their businesses, often unknowingly violating regulations, which can lead to fines or even market exclusion. To solve this problem, Lumura was developed to provide real-time, accurate answers to tax-related questions, helping business owners stay compliant and avoid costly mistakes. The chat agent is integrated with Pinecone for efficient data embedding, utilizing Langchain as the LLM framework and OpenAI for API functionality. The system leverages publicly available tax data from CAC, which is processed and embedded to ensure the chatbot can provide accurate, up-to-date answers on relevant tax laws. With Lumura, business owners can interact with the system to clarify tax regulations specific to their industry, ensuring they remain within compliance and safeguard their businesses from unnecessary fines or shutdowns.

EduPeerHub is a platform designed to help students prepare for their exams by providing easy and accessible resources, especially for those without access to traditional materials. The goal of the platform is to ensure that all students, regardless of their resources, can effectively study and succeed in their exams. As a Cloud Engineer on this project, I was responsible for deploying the entire infrastructure using AWS. This included setting up and automating CI/CD pipelines, registering the platform’s domain, ensuring secure certifications, and seamlessly linking the frontend and backend systems. Additionally, I set up a proxy with Nginx, ensuring efficient traffic management and improving the platform’s scalability and performance. While the platform is currently operational, its availability may change over time.

In this project involved deploying an e-commerce webpage for a retail store, utilizing a microservices architecture hosted on Kubernetes. The goal was to create a scalable, high-availability solution that could handle multiple services and ensure seamless user experiences for customers. I used Terraform for infrastructure as code to automate and manage the entire deployment. This included setting up Amazon RDS for database management and configuring Amazon EKS to manage the Kubernetes cluster. I made the service live by integrating Elastic Load Balancer (ELB), ensuring efficient traffic distribution across multiple instances for high availability. Additionally, I configured network settings, setting up private and public subnets to ensure proper access control and security for the services. This setup ensured smooth access for both internal and external traffic while maintaining optimal security.