At Data Axle, I architected an event-driven email pipeline using Apache Kafka and AWS EventBridge, migrating from SparkPost to AWS SES and cutting recurring SaaS costs by 40%. I also engineered an SMS orchestration service with Genesys Cloud that processes 450K+ daily dispatches with 99.98% delivery reliability.
I refactored the Datalynx data access layer with Redis distributed caching and connection pooling, reducing p99 API latency by 60% during peak load. I also built low-latency microservices for the Buyer-Intent engine, improving lead conversion accuracy by 25%.
In personal projects, I fine-tuned Llama-3.2-3B for Amazon product price prediction, cutting prediction error by 41% against the base model. I also built MedRAG Pro, a medical assistant RAG pipeline that improved retrieval accuracy by 35% and added evaluation and PII guardrails.
