Alan Sammarone

We developed an end-to-end system for bank transaction classification using weak supervision and deep learning.

I was involved in the ideation and research phases and led the productionization effort. My work included designing and implementing the real-time inference architecture, integrating the system with Kafka and KSQL for event batching, and deploying scalable models via Kubernetes and Kubeflow Pipelines.

The system uses heuristics and unsupervised embeddings to generate weak labels, which are then used to train GRU-based discriminative models. Our pipeline outperformed Plaid API in several tasks, achieving over 90% accuracy across nine financial categories. The architecture was optimized for rapidly onboarding new classification tasks with minimal manual effort.

The project is a Python back end for an MVP that delivers an interactive voice interface capable of handling end-to-end customer support tasks without human intervention. It leverages three OpenAI capabilities in sequence: gpt-4o-transcribe to convert speech to text, gpt-4o (with tool-calling support) to process the input and generate a response, and gpt-4o-mini-tts to convert the response back to speech.