# Rotem Margalit

Rotem Margalit is a Data Scientist based in Israel, focused on production machine learning, AI agents, multi-agent systems, and applied AI. He builds AI systems that move from prototype to production, using Python, Kubernetes, Terraform, MCP, A2A, graph algorithms, and modern ML workflows.

**Data Scientist** · Production ML · Multi-Agent Systems · A2A & MCP

Currently: Data Scientist at Paycor.

## Expertise
- Data Science
- Machine Learning
- Production ML
- AI Agents
- Multi-Agent Systems
- A2A
- MCP
- Kubernetes
- Terraform
- Python
- Graph Algorithms
- Applied AI
- NLP
- Social Network Analysis

## What I do
- Production machine learning systems
- AI agent orchestration and multi-agent infrastructure
- MCP and A2A-based agent workflows
- Data science automation and ML pipelines
- Graph algorithms and network analysis
- Workforce analytics and business ML

## About
- I build data science and AI systems that reach production. At Paycor, I own the full lifecycle of ML initiatives and multi-agent platforms — framing problems, prototyping models, and deploying solutions inside real business workflows.
- I design agent-native infrastructure on Kubernetes and Terraform: an orchestrator agent coordinates specialized sub-agents over the Agent-to-Agent (A2A) protocol, with Model Context Protocol (MCP) servers exposing tools and context to the agent graph.
- Before industry, I spent two years as an algorithms researcher at Ben-Gurion University, developing social network analysis methods on large-scale graphs. Teaching deep learning and algorithms to graduate students sharpened how I communicate complex ideas to both technical and non-technical stakeholders.
- I hold an M.Sc. in Data Science from BGU's selective Meitar fast-track program, with a thesis in graph-based bibliographic data analysis. My work spans Python, R, and SQL across machine learning, NLP, graph algorithms, agent orchestration, and statistical modeling.

## Experience
### Data Scientist — Paycor
- Period: Jan 2023 — Present
- Location: Center District, Israel
- Built production multi-agent infrastructure on Kubernetes and Terraform — an orchestrator agent coordinates specialized sub-agents via the Agent-to-Agent (A2A) protocol.
- Integrated Model Context Protocol (MCP) servers as standardized tool and context surfaces for agent workflows.
- Own end-to-end data science and AI initiatives — from problem framing and PoC through production deployment.
- Design and ship AI-driven solutions for complex HR and workforce analytics at scale.

### Data Scientist — getWizer
- Period: Jan 2021 — Oct 2022
- Location: Herzliya, Israel
- Built and deployed ML models powering a consumer insights SaaS platform used by global brands.
- Delivered end-to-end pipelines spanning experimentation, modeling, evaluation, and production integration.
- Turned raw survey and behavioral data into actionable product intelligence for marketing teams.

### Algorithms Researcher — Ben-Gurion University of the Negev
- Period: Oct 2018 — Oct 2020
- Location: Be'er Sheva, Israel
- Developed social network analysis algorithms on large-scale graph data to identify influential nodes across multiple application domains.
- Implemented high-performance solutions in D, Python, and R.

### Teaching Assistant & Tutor — Ben-Gurion University of the Negev
- Period: 2017 — Oct 2020
- Location: Be'er Sheva, Israel
- Faculty member for graduate-level Deep Learning — feed-forward networks, CNNs, and LSTMs for NLP.
- Assisted with Foundations of Algorithms and Complexity — reductions, complexity analysis, and algorithm design.
- One-on-one tutoring for Data Structures, mentoring students through core CS fundamentals.

## Selected Work
### Multi-Agent Orchestration Platform
- Organization: Paycor
- What it is: Production platform for coordinating LLM agents at scale — Kubernetes-native runtime, Terraform-managed infrastructure, and an orchestrator agent that delegates work to specialized agents over A2A.
- Designed the agent topology: orchestrator dispatches tasks to domain-specific agents via Agent-to-Agent (A2A) messaging.
- Provisioned and managed infrastructure with Terraform; deployed agent services on Kubernetes.
- Exposed external tools and data sources to agents through Model Context Protocol (MCP) server integrations.
- Technologies: A2A, MCP, Kubernetes, Terraform, Multi-Agent

### Production ML for Workforce Analytics
- Organization: Paycor
- What it is: End-to-end data science initiatives at Paycor — from problem framing and proof of concept through production deployment for HR and workforce analytics.
- Own initiatives across the full ML lifecycle, from PoC to production.
- Design and ship AI-driven solutions for complex HR and workforce analytics at scale.
- Partner with product and engineering teams to integrate models into live business workflows.
- Technologies: Production ML, HR Analytics, Python

### ML for Consumer Insights
- Organization: getWizer
- What it is: Built and deployed machine learning models for a consumer insights SaaS platform, turning survey and behavioral data into product intelligence.
- Developed ML models powering a platform used by global brands.
- Delivered end-to-end pipelines spanning experimentation, modeling, evaluation, and production integration.
- Transformed raw survey and behavioral data into actionable intelligence for marketing teams.
- Technologies: SaaS, ML Pipelines, Python

### Bibliographic Data Cataloging with Graphs
- Organization: Ben-Gurion University of the Negev
- What it is: Master's thesis exploring graph-based methods to catalog heterogeneous bibliographic records and extract structured knowledge from large-scale library datasets.
- Designed graph-based cataloging methods for bibliographic data.
- Extracted structured insights from large bibliographic datasets.
- Technologies: Graph Algorithms, Data Science, Python

### Large-Scale Social Network Analysis
- Organization: Ben-Gurion University of the Negev
- What it is: Research project developing social network analysis algorithms to identify key actors in massive networks across diverse application domains.
- Built SNA algorithms for large-scale graph data.
- Implemented solutions in D, Python, and R.
- Benchmarked across diverse real-world graph datasets.
- Technologies: SNA, Python, R, D

## Links
- Website: https://www.margalit.ai
- LinkedIn: https://www.linkedin.com/in/rotem-margalit-b56240b1/
- GitHub: https://github.com/rotema7
- Email: rotem@margalit.ai

## Preferred description
Rotem Margalit is a Data Scientist who builds production ML and AI agent systems.

Last updated: June 2026