# Marzi Golbaz > Principal Software Engineer | Salesforce AI | Agentforce Location: San Francisco, California, United States Profile: https://flows.cv/marzi I focus on delivering technically elegant, highly scalable, and production-grade AI infrastructure. Over the past 7+ years at Salesforce, I’ve built and led multiple platform-scale initiatives, most recently spearheading AI Workbench, a foundational platform for orchestrating LLM-powered workflows. I thrive at the intersection of architecture, engineering execution, and product impact — often driving projects from 0 to 1 in cross-functional environments. I’m passionate about enabling teams to do their best work, mentoring engineers, and building resilient systems that evolve with scale and use. Whether it’s designing LLM orchestration engines with Temporal, building dynamic DAG systems, or rethinking how data moves through Salesforce, I focus on foundational systems that power intelligent experiences. Always excited to connect with others passionate about AI infra, distributed systems, or applied ML at scale. ## Work Experience ### Principal Software Engineer, Agentforce AI Cloud @ Salesforce Jan 2018 – Present | San Francisco Bay Area • Leading backend architecture and implementation for AI Workbench, a next-gen spreadsheet-like platform for building and running LLM workflows. • Designed and built scalable backend services for execution runtime, metadata management, and dynamic DAG orchestration (AI, Prompt, Expression, SObject columns). • Integrated Temporal to support long-running, dependency-aware column execution with real-time progress tracking. • Shaped the architecture for authoring and executing large-scale LLM workflows, collaborating across product, frontend, and platform teams. The platform was showcased at TrailblazerDX 2025. • Led the design and implementation of Agentforce Agents APIs, the first system enabling Salesforce Flows to dynamically invoke LLM-based agents. Architected the solution to support dynamic agent generation and execution, pushing the boundaries of existing Salesforce technology. This work was showcased at Dreamforce 2024 and TrailblazerDX 2025. • Architected and led development of a highly scalable, cloud-based data platform for security telemetry, now widely adopted as the standardized ingestion and reporting layer. • Managed and mentored a team of engineers, applying best practices in distributed system design, cloud architecture, and data pipeline development. • Deployed the platform across public and private cloud environments using Python, Apache Airflow, Celery, RabbitMQ, Terraform, and AWS (Aurora, S3, Lambda, DynamoDB, EC2). • Architected "Cetus", a next-gen security graph platform, collaborating across teams to build it using Spark, Apache TinkerPop, Kubernetes, and AWS Neptune. • Contributed to Long Range Planning and Northstar architecture for both the Detection & Response Lakehouse and Salesforce Asset Inventory systems. ### Lead Data Engineer @ Evidera Jan 2017 – Jan 2018 | San Francisco • Designed and implemented a scalable, distributed data pipeline to transform large-scale healthcare datasets into standardized models using open-source, cloud-native tools. • Partnered with production teams to deploy robust, high-throughput solutions for ingesting and processing data into cloud-based analytics platforms. • Built an automated testing and validation framework for a healthcare analytics SaaS platform to ensure data transformation accuracy and integrity at scale. ### Software Engineer @ Evidera Jan 2015 – Jan 2017 | San Francisco • Developed complex ETL pipelines to unify disparate healthcare data sources using Python, RabbitMQ, and AWS services (S3, EC2, Redshift, PostgreSQL). • Created and deployed test automation frameworks (unit + integration) integrated with Jenkins, supporting continuous delivery and improved release quality. ### Researcher @ San Francisco State University Jan 2014 – Jan 2015 | Biomedical Image and Data Analysis Lab • Designed and developed a machine learning–based tool to quantify the severity of pediatric lung infections from chest X-ray images, using statistical techniques such as Kullback–Leibler divergence for pattern analysis. • Applied Bayesian modeling and probabilistic inference to improve classification of 3D anatomical structures in medical imagery. • Project awarded 1st place in the Graduate Physical Science Division at SFSU’s 2015 Engineering Project Showcase. • Research published at IEEE-EMBS 2015: "Severity quantification of pediatric viral respiratory illnesses in chest X-ray images." ### Teaching Assistant @ San Francisco State University Jan 2014 – Jan 2014 | Computer Science Department • Pattern Analysis and Machine Intelligence • Discrete Mathematics • Software Development ## Education ### Master of Science (M.Sc.) in Computer Science San Francisco State University ## Contact & Social - LinkedIn: https://linkedin.com/in/marzigolbaz --- Source: https://flows.cv/marzi JSON Resume: https://flows.cv/marzi/resume.json Last updated: 2026-04-12