Ben Labaschin
Principal Machine Learning Engineer
Building intelligent systems that deliver measurable impact—from pioneering enterprise-scale GenAI platforms to developing ML systems that drive millions in business value.
Download Resume
Overview
I'm a Principal Machine Learning Engineer at Workhelix, where I've been building enterprise-scale GenAI platforms and production ML systems as a founding engineer. I've written two O'Reilly books—"What Are AI Agents?" and "Managing Memory for AI Agents"—covering the practical considerations of building and deploying AI agent systems. I recently published research in AEA Papers and Proceedings on measuring firm-level exposure to large language models and their potential productivity impacts. I spend most of my time figuring out how to actually deploy AI systems that work reliably in production—from async LLM APIs and embedding systems to the messy real-world challenges of putting GenAI into enterprise workflows. My work spans the full stack, and I split my interests between the theory of AI and its impacts on labor, to deploying production grade LLMs and AI Agents.
AI Systems Architect
Building scalable ML infrastructure from the ground up
Innovation Driver
2 patents and multiple business-critical ML innovations
Value Creator
Delivered $8M+ in savings and helped secure $96M investment
Researcher & Author
See recent research on firm-level exposure to LLMs in AEA Papers and Proceedings, and see O'Reilly series on AI Agents
Experience
Principal Machine Learning Engineer
- Founding Engineer: Leads core platform development at Workhelix, parallelizing highly scalable async LLM APIs and custom SOTA embedding systems, building and maintaining our internal LLM and agent deployment platform for Fortune 50 enterprise customers.
- ML/GenAI Analytics Platform: Architected multi-container Docker system for real-time data embedding and workflow analysis, combining proprietary "task" classification algorithms with predictive models (PyTorch, FastAPI, AWS ECR/ECS).
- Analytics & Insights: Developed novel complexity metrics to quantify and forecast GenAI's impact on engineering workflows, enabling data-driven insights on productivity for enterprise customers.
- API Architecture & Data Pipeline Engineering: Engineered high-throughput GitHub/GitLab extraction system with asyncio rate limiting and GraphQL cursor pagination, building robust task queues/concurrency controls that reduced repository processing time from 30 to 2 minutes.
- Seed to Series A Growth: Drove critical technical development that enabled Workhelix's growth from seed to Series A ($75M valuation), attracting investment from AI leaders including Andrew Ng, Mira Murati, and Yann LeCun.
Senior Data Scientist
- Strategic Partnership Leadership: Led ML engineering for Capital One Travel partnership, creating foundational ML systems that helped secure a $96M investment and drove Hopper Cloud to 40% of company revenue.
- Technical Architecture: Designed and implemented multi-tenant ML systems including multi-arm bandits and price freeze algorithms using GCP, enabling secure and scalable travel booking solutions for enterprise partners.
- MLOps Innovation: Spearheaded DevOps/MLOps implementation for Hopper Cloud, establishing CI/CD pipelines and orchestration workflows with GitHub Actions and Kubeflow to support rapid scaling of the B2B platform.
Data Scientist
- Cost Optimization & ML Systems: Delivered $8M in savings through optimized shipment prioritization and automated pricing systems, a result from spearheading A/B testing and ML initiatives using XGBoost.
Data Scientist
- Investment Analytics: Led A/B testing with power analysis and regression modeling to optimize real estate investment decisions, driving multi-million dollar property savings.
- Analytics Engineering: Developed Python analytics pipeline (scipy, scikit-learn) for apartment renovation ROI analysis and geodata-based warehouse investment models.
Economist Researcher / Data Scientist
- Risk Innovation: Pioneered telematics-based risk modeling for shared mobility companies, resulting in two patent applications and creating a new business vertical.
- Innovation Recognition: Won 2019 Hackathon for AWS-based NLP modeling, with research featured in Business Insider.
Adjunct Lecturer
- Developed and taught Python-based AI/ML curriculum to 30+ students, bridging academic concepts with industry applications.
Skills
Programming
AI/ML
Distributed Systems
Cloud
Deployment
Databases
Publications & Talks
Building Stateful AI Agents
ODSC West (Talk)
Managing Memory for AI Agents
O'Reilly Media (Book)
Extending "GPTs Are GPTs" to Firms
AEA Papers and Proceedings
Building With AI: How I Build Quick POCs with LLMs
Wharton Guest Lecture
A Normie Approach to Validating LLM Outputs
AI.Science Talk
What Are AI Agents?
O'Reilly Media (Book)
Building an HTTPS Model API for Cheap: AWS, Docker, and the Normconf API
Talk
Patents
Shared Mobility Simulation and Prediction System
USPTO 20190347941
Matching Drivers With Shared Vehicles To Optimize Shared Vehicle Services
USPTO 20190347582