Hello, I am Joyce. With 8+ years experience in Software Engineering across DevOps, Machine Learning, and MLOps. I specialize in designing production-grade ML systems and embedding AI into real-world applications.
I enjoy building scalable ML and LLM pipelines, integrating observability, and automating the full lifecycle of model deployment. I am passionate about delivering secure, private AI solutions that bring measurable business value.
I also work with these Languages and Technologies:
I massage data and design intelligent systems.
I build and deploy ML, Deep Learning, and Generative AI solutions using tools like SageMaker, MLflow, and LangChain. My work includes implementing recommendation engines, LLM-powered chatbots, and predictive analytics pipelines in production environments.
I design end-to-end MLOps workflows that automate data ingestion, model training, deployment, and monitoring.
My experience includes building pipelines with Airflow, MLflow, and Kubernetes, and deploying models securely in AWS and GCP.
I integrate observability with Prometheus, Grafana, and ElasticSearch to ensure reliable operations.
I engineer production-grade Generative AI solutions combining LLMs, vector databases, and scalable APIs.
My projects include Retrieval-Augmented Generation systems, document intelligence platforms, and secure ML services.
- Led the architecture and delivery of production Retrieval-Augmented Generation (RAG) pipelines using LangChain and pgvector across AWS and Azure, supporting over 15 internal AI applications.
- Developed scalable ML workflows with Airflow, MLflow, Docker, and Kubernetes, reducing model retraining time by 40% and improving deployment frequency by 3x.
- Built and optimized GPT-3/4 and BERT-powered services for text summarization and classification, increasing content processing throughput by 60%.
- Established CI/CD pipelines with GitHub Actions and MLflow, achieving 100% reproducible deployments and traceable model lineage.
- Mentored a cross-functional team of 5 engineers, driving adoption of MLOps best practices and stakeholder collaborations.
- Designed and productionized a deep learning model for GNSS signal interference detection using CNNs on GPUs, improving detection accuracy by 35% compared to legacy systems.
- Granted US Patent #11988753 for GNSS-receiver interference detection using deep learning.
- Delivered observability dashboards with Prometheus, Grafana, ElasticSearch, and MySQL, reducing incident resolution times by 60%.
- Led Python codebase modernization and maintained comprehensive technical documentation for multiple products.
- Reviewed and merged 300+ pull requests, driving continuous improvement and mentoring junior developers.
- Designed and delivered end-to-end ML solutions for over 12 clients in e-commerce, real estate, and fintech, driving business KPIs such as lead conversion and customer retention.
- Built and deployed recommendation engines and predictive analytics pipelines using AWS SageMaker, Airflow, and Kubernetes, reducing time-to-insight by 50%.
- Developed Retrieval-Augmented Generation systems leveraging OpenAI APIs, LangChain, and pgvector to power custom search and conversational AI experiences.
- Implemented monitoring and observability frameworks with Prometheus and Grafana to track model drift, latency, and performance metrics.
- Created reusable infrastructure-as-code templates and CI/CD workflows to accelerate project delivery and ensure reproducibility.
- Developed Python and Java components to extend simulation platform capabilities and improve system reliability.
- Automated testing workflows using Selenium and Jenkins, reducing regression cycles by 40% and increasing release confidence.
- Built and maintained CI pipelines to streamline code integration and deployment processes.
- Diagnosed and resolved software defects, contributing to higher-quality releases and reduced support incidents.
Interactive Intelligent Systems specialization with coursework in AI, Machine Learning, Health Informatics, and Algorithm Design.
