Brent Goldstein

About Brent Goldstein

Currently a Machine Learning Engineer and Data Scientist at ADP.May 2024 RPI CS grad. A passionate programmer specializing in automated unit testing and framework development, machine learning, game development, and AI, with a track record of success at Warner Bros. Games Central Game Technology. Contributed to Curtail\'s architecture, seamlessly integrating gRPC and HTTP2. Actively engaged in the RPI Game Jam club. Seasoned game designer and developer, with contracting experience at Amplifier Game Invest. Proficient in Unity, Unreal, and Godot. Founder of B-Games, a sole proprietorship focused on game and software development.Website: Itchi.io

Experience

1. Associate Machine Learning Engineer and Data Scientist

   ADP

   Jul 2024 — Present · Essex County, NJ, US

   Revamped the Databricks ADP AI Playground, managing access and full model lifecycle (registration, versioning, endpoints, cross‑workspace orchestration) for 20+ active models- Automated enterprise model registration and CI/CD pipelines, cutting registration/deployment time by >40% and supporting 100+ biweekly production releases across teams- Built ADP AI Playground dashboard, tracking prompt volume, token costs, guardrail events, and error patterns across US/EU workspaces for over 500 users- Led a team of engineers to deliver ADP’s first secure RAG system, supporting user file uploads with automated PII scanning, sanitized storage, and retrieval workflows that provide users with full document and session history while AI governance- Designed the full RAG retrieval pipeline’s chunking and embedding workflows, vector indexing, retrieval logic, and generation integration, also implementing support for multiple embedding models and concurrent query sessions, enabling near real‑time, natural‑language retrieval across multiple instances- Cut RAG retrieval latency from minutes to less than 3 seconds by optimizing index refresh and retrieval paths- Engineered a secure modular RAG frontend and Flask API layer, integrating multi‑APIs to power full‑stack RAG and agent workflows- Scaled AI observability from POC to production with SHAP explainability, drift/bias detection, custom metrics, and OTEL logging for generative AI, AI Agents, and Machine Learning workloads- Built synthetic datasets and reverse‑mapping tools to trace outputs to inputs for privacy‑preserving validation and targeted test cases, creating a tool with over 95% accuracy to the authentic data- Deployed fully air-gapped platforms with Docker, Terraform, EKS, Helm, S3, IAM and implemented VPC/EKS separation, image transfer flows, and OpsGenie alerts- Migrated 300+ models into Unity Catalog and automated promotion workflows to improve governance and accelerate deployment cycles.

Education

• Rensselaer Polytechnic Institute

  Bachelor's degree, Computer Science

  2020 — 2024