Roran Brian

Developed high-throughput AI inference gateways using Go, gRPC, and Protobuf for real-time decision-making and customer interaction analytics. Built NLP-driven RAG models with pgvector and Redis for document retrieval. Created real-time AI feature extraction pipelines (Kafka → Go consumers → S3/Parquet), reducing training data freshness from 24 hours to 5 minutes. Implemented transformer-based models (BERT, T5) with HuggingFace transformers and deployed via TensorFlow Serving and KServe. Led AI/ML model fine-tuning with LoRA and CUDA containers, achieving a +21% performance boost for GPT-4 fine-tuning. Developed data pipelines for Generative AI and Text-to-Image models (Stable Diffusion, CLIP). Deployed NLP-based agents with open-source LLMs and integrated with chatbot frameworks; added Prometheus, Grafana, and Datadog APM for real-time monitoring.

Led AI/ML and data engineering initiatives, building scalable data pipelines and real-time models to enhance inventory and order management systems. Implemented batch and stream ETL in Go and Apache Kafka, applying NLP models for extraction and classification from unstructured data sources. Architected AI-powered services with spaCy, transformers (Hugging Face), and TensorFlow for customer interaction analysis. Built high-performance Go microservices for AI inference, reducing latency by 35% for predictive analytics. Integrated LLMs (GPT-3/4, BERT) for real-time text classification and sentiment analysis of customer feedback. Orchestrated deployments with AWS CDK (Python) and served TensorFlow models via TensorFlow Serving in cloud-native environments. Optimized pipelines with Dask and Ray; integrated PostgreSQL and DynamoDB for multi-source data storage. Implemented CI/CD with Argo CD, Kubernetes, and GitHub Actions; secured endpoints with OAuth 2.0.

Managed the development of a Python microservice for inventory and order management across 2,700 oil & gas stations. Implemented a delay-and-retry approach for 3PL integrations; architected AWS CDK-based infrastructure (Lambda, EC2, S3, RDS, SNS, SQS); hosted Docker images on AWS EKS with ArgoCD; used gRPC with Protobuf for inter-service communication and integrated DynamoDB and PostgreSQL. Streamlined CI/CD with GitHub Actions; led Python API development (Python 3.9); implemented OAuth 2.0 authentication, achieving >85% test coverage with pytest/unittest. Integrated Stripe/PayPal in Billing; migrated a Social Management Service to Flask/Django; maintained a React frontend; set up OpsGenie alerts and Datadog metrics.