Open to opportunities

Benny Franco

@bennyfranco

Message

Data Architect building production-grade AI and GenAI systems that drive measurable business gains.

United States

Message

What I'm looking for

I’m looking to architect and productionize AI and Generative AI systems end-to-end—data foundations, Production MLOps, and measurable business outcomes—where transparent evaluation, auditable reasoning, and scalable pipelines are valued.

I’m a Data Architect with 8+ years spent building AI engines for enterprise platforms and consumer marketplaces. I focus on end-to-end delivery across Data Architecture, Production MLOps, and strategic Generative AI—turning complex data and model workflows into decision systems that teams can trust and scale.

Most recently, I devised a novel multi-agent “Composition of Experts” RAG system for a Tier-1 investment bank, unifying live financial data and proprietary research into a single decision engine. I built a low-latency data fabric powering both SQL-querying and semantic search agents, enabling wealth managers to generate client reports 80% faster and improving trade analysis accuracy by 35%.

I also architected revenue-impacting AI at scale, from Styleseat’s real-time Kafka/Flink dynamic pricing engine (15% average revenue lift; 20% fewer unfilled appointments) to Sweet Rush’s 85%-accurate “Learner Success” model and an AWS SageMaker prescriptive churn prevention system that drove a 50% increase in marketing ROI. I bring a production mindset—tool-use optimization, auditable reasoning traces, evaluation/benchmarking rigor, and operational feedback loops—so AI creates durable business value.

Experience

Work history, roles, and key accomplishments

Current

Data Architect

Current

Samba Nova

Aug 2024 - Present (1 year 10 months)

Devised a multi-agent “Composition of Experts” RAG system for a Tier-1 investment bank, unifying live financial data and proprietary research into a decision engine. Built a low-latency data fabric that enabled wealth managers to generate client reports 80% faster and improved trade analysis accuracy by 35%, and delivered auditable, tool-using agent workflows that supported the company’s flagship

Retrieval Augmented Generation (RAG)LangGraph LlamaIndex Vector Databases Data Engineering SQL Semantic Search

Senior Data & AI Engineer

Styleseat

Jul 2022 - Aug 2024 (2 years 1 month)

Engineered an AI-powered dynamic pricing engine using real-time Kafka and Flink pipelines to process booking trends and local demand signals for revenue optimization. Developed an LSTM forecasting model adopted by 50,000+ stylists, increasing average revenue by 15% and reducing unfilled appointment slots by 20%, while improving forecast reliability with interpretable decision trees and structured-

Apache Kafka Apache Flink Feature Engineering Machine Learning Model Interpretability

Data Scientist

Sweet Rush

Jan 2022 - Jun 2022 (5 months)

Developed a predictive “Learner Success” model to identify high-potential leaders for a Fortune 500 client using hundreds of engineered learner activity signals. Correlated engagement data with on-the-job performance KPIs to accurately identify future top-performing managers with 85% accuracy, and launched the model as a new high-margin analytics product validating the long-term business impact of

predictive modeling Feature Engineering Python Pandas SQL KPI Analysis Model Evaluation Analytics

AI Engineer

Clustera

Feb 2020 - Dec 2021 (1 year 10 months)

Built real-time pipelines ingesting contact center data from streaming sources, APIs, and distributed caches to create a unified analytics data layer. Developed transformation and feature engineering workflows that turned call transcripts and interaction data into ML-ready datasets for sentiment analysis, churn prediction, and call volume forecasting, enabling near real-time operational insights f

ETL ELT Apache Kafka AWS Kinesis Data Warehousing Feature Engineering Sentiment Analysis

ML Engineer

Etyon

Nov 2017 - Jan 2020 (2 years 2 months)

Built an enterprise-scale prescriptive churn prevention system on AWS SageMaker that became the foundation of the client retention strategy. Used uplift modeling to move from reactive to ROI-optimized interventions, increasing marketing ROI by 50% and improving retention by reducing churn an additional 10% versus the prior model.