Liam Martin Pinson
Software Engineer
I build robust, scalable systems and craft seamless developer and end-user experiences. Passionate about cloud-native architecture, CI/CD automation, AI/ML integrations, and transforming complex infrastructure challenges into elegant, efficient solutions.
Where I've worked
A timeline of my professional journey building software and infrastructure at scale.
Oracle Cloud (OCI) Integration Engineer
AG Jeans
Designed and implemented end-to-end integration workflows between PolyPM, Oracle JDE, and MS SQL Server using Python on Oracle Cloud Infrastructure, ensuring automated data synchronization for supply chain operations.
- Architected and deployed Python backend services replacing 20+ legacy integration pipelines, improving system reliability, reducing operational overhead, and enabling independent service scaling
- Built and operated automated CI/CD pipelines (GitLab, OCI) for production backend services, enabling zero-downtime deployments and taking full operational ownership including incident response
- Containerized Python backend microservices with Docker and led architecture planning for migration to Kubernetes (OKE), designing for horizontal scalability and service isolation
- Designed and operated distributed data synchronization services using REST APIs and integration middleware, with structured logging, monitoring, and reliability-focused service workflows across enterprise backend systems
- Managed cloud-native backend workloads on OCI including VM provisioning, scheduled automation, and service lifecycle management in production environments
Application Developer Intern — SAP Transportation Management
Yamaha Motor Corporation, USA
Led infrastructure modernization initiatives and managed cloud platforms for high-traffic applications.
- Integrated a visibility provider into SAP ECC and SAP TM to monitor outbound truck freight via RESTful API through SAP BTP.
- Developed RESTful APIs exposed via BTP-IS as middleware to process shipment tracking events from visibility providers back to SAP ECC, storing data in custom tables consumed by the Dealer Portal Webdynpro application.
- Created 100+ LeanIX interface diagrams documenting all application relationships within S/4HANA and mapping current (SAP ERP) to future (SAP S/4HANA) outbound/inbound shipment processes.
Things I've built
A collection of projects spanning web applications, infrastructure tooling, and developer experience.
Spotify Hit Predictor
Spotify Hit Predictor is a full-stack machine learning web application that predicts whether a Spotify track has the potential to become a hit by analyzing its audio characteristics. Built with Next.js and Tailwind CSS on the frontend and FastAPI on the backend, the application integrates the Spotify API to search and retrieve track metadata, uses spotDL to download audio files, and leverages Librosa to extract over 20 acoustic features including tempo, energy, spectral centroid, and MFCCs. A trained PyTorch neural network processes these features to generate a hit probability score, while Google Gemini AI produces natural language producer insights explaining the prediction. The entire system is containerized with Docker and deployed to Google Kubernetes Engine (GKE) on Google Cloud Platform via a GitLab/GitHub CI/CD pipeline, demonstrating proficiency in full-stack development, deep learning, audio signal processing, cloud infrastructure, and DevOps practices.
Stock Portfolio Agent
This project is a production-focused Stock Agentic AI platform that combines a React/Next.js monitoring dashboard with a polyglot backend built on Java/SpringBoot for API and portfolio services and Python/FastAPI plus Python/PyTorch for agent orchestration and machine learning. It uses an MCP-based multi-agent architecture where a Research Agent gathers market context, a Trading Agent generates and validates trade decisions, and a Summary Agent produces updates and performance insights. The intelligence layer integrates Gemini AI reasoning, RAG over pgvector for financial knowledge retrieval, and a Temporal Fusion Transformer model for short-term market forecasting, while execution flows through Alpaca API paper trading for safe end-to-end validation. The platform is containerized with Docker and deployed on GKE with CI/CD via Cloud Build and GitLab deployment pipelines, demonstrating strong skills in multi-agent AI systems, distributed backend development, ML time-series modeling, RAG integration, cloud-native DevOps, and financial application engineering.
Portfolio Website
A performant, SEO-optimized portfolio website showcasing professional experience and projects. Built with Next.js 14, TypeScript, and Tailwind CSS with smooth animations and dark mode support.
Technologies & tools
The languages, frameworks, and platforms I work with every day.
Languages
Frontend
Backend
DevOps & Cloud
Tools & Practices
Let's connect
Have an interesting project or opportunity? I'd love to hear from you. Drop me an email or connect on social media.
liammartinpinson@gmail.com