Marcos Keyser

About Marcos Keyser

I bridge the gap between \"Vibe Checks\" and production-grade performance. As a Principal Applied Research Scientist, my role is to ensure Generative AI products—from RAG systems to agentic workflows—actually work in high-stakes, domain-specific environments. My core thesis is simple: NEUROSYMBOLIC AI IS THE ONLY PATH TO RELIABLE GEN AI.The industry is moving past the experimental phase where \"chat\" was enough. Today, the challenge isn\'t whether AI can perform a task, but whether it can explain, defend, and verify its decisions. I focus on building this \"System 2\" reasoning—grounding the fluid pattern recognition of LLMs with the deterministic safety and accountability of symbolic logic. CORE FOCUS AREAS: AUTONOMOUS RELIABILITY: Engineering systems that move beyond probabilistic guessing. I integrate LLMs with formal verification and logic engines to ensure mission-critical workflows are both accurate and auditable. PERFORMANCE ARCHITECTURE: Serving as the bridge between R&D and product. I specialize in moving agents from \"good enough for a demo\" to \"safe for production\" in regulated industries where mistakes have real-world consequences. DOMAIN-SPECIFIC REASONING: Applying distinct verification strategies for complex data archetypes—procedural, declarative, narrative, and argumentative—ensuring the AI respects the underlying \"world model\" of the business. EXPLAINABLE AGENCY: Designing for transparency. I build agents that can provide traceable reasoning paths, enabling true human-in-the-loop governance for enterprise-scale deployments.I turn research into institutional infrastructure, transforming raw probabilistic models into robust, reasoning-capable agents.

Experience

1. Principal Applied Research Scientist Neurosymbolic Ai & Performance Architecture

   New York Life

   Jul 2017 — Present · New York, NY, US

Skills

• Natural Language Processing
• Sas
• R
• Predictive Modeling
• Actuarial Science
• Machine Learning
• Stadistical Analysis
• Risk Management
• Python
• Data Science
• Deeep Learning
• Social Media Data Mining
• Data Mining
• Sql
• Social Network Analysis
• Web Scraping
• Spatial Analysis