Machine Translation Experience (LLMs)

• Designed and implemented smart diffusion image text replacement

• Processed and created quality scores of LLMs to determine best provider per use case

• Created metric dashboards and airflow SQL pipelines to automate quality reports and subsequent actions

• Created and assigned terraform AWS security roles

• Ground-up developed a scalable service for high client throughput at sub-millisecond latency (Golang)

• Achieved 98% reduction in client-facing latency from 400ms to 5ms by optimizing Redis cache design

• Designed and developed LLM provider integration (text-bison, palm2, gemini, google, chatGPT-4o)

• Prompt engineered templates to yield high quality and reliable specific results

• Designed and created multiple new services from the ground up complete with build pipelines and dashboards

• Developed algorithm to suggest market-specific content based on local user trends

• Designed feed that fosters new user interests by presenting known liked content near new themes (Python, Go)

• Conducted multi-variable experiment to test new feed algorithm that increased user engagement by 20%

• Built end-to-end pipeline for low-latency content delivery from Airflow to Redis, reacting to real time trends

• Developed machine learning model to predict user intent when calling support with a 70% accuracy rate

• Processed large datasets (+50M rows) using Hive DB and Apache Spark data pipeline

• Analyzed data and identified optimal parameters for use with a Natural Language Processing (NLP) model

• Implemented model in a production environment using Python and Spark, resulting in a 30% cost savings