Summary
Lead Data Scientist with 6+ years building agentic AI systems and decision intelligence platforms. Specialize in deploying production-scale solutions using LangGraph, RAG, and LLMs that have automated $80B+ in business processes and freed 70+ FTEs. Track record of turning complex AI into practical tools that drive ROI. Strong in cloud deployments (GCP, AWS), Kubernetes, and production ML systems.
Experience
Agentic AI & Autonomous Decision Systems
- Built enterprise agentic AI platform using LangGraph/LangSmith that automatically generates pricing bids from unstructured sources (PDFs, emails, Excel files), handling $80B+ in annual freight bids. Cut turnaround time by 40 hours and freed up 20 FTEs
- Created MCP server infrastructure that lets agent systems pull costing guidance and deliver consistent, real-time decision support to commercial reps
- Deployed conversational analytics agents that give business users on-demand pricing intelligence and recommendations
Production ML & Decision Intelligence
- Built and deployed costing engines for all transactional and contractual freight—the backbone of our pricing strategy
- Rewrote critical time-series library (farseer) in Rust for better performance in production forecasting pipelines
- Developed predictive models and NLP systems for demand forecasting, route optimization, and customer behavior analysis
Cloud Infrastructure & MLOps
- Deployed and maintained AI systems in Kubernetes using Helm, troubleshot OOM errors and connection issues with k8s/k9s
- Built AI governance frameworks around "trust but verify" to handle model overconfidence in business-critical systems
- Set up CI/CD pipelines for model deployment, monitoring, and observability
- Built automation platform using webhooks that captured real-time events and automated task management across IBM Finance departments, saving 50+ FTE equivalents
- Developed time-series forecasting package in R and mentored junior data scientists
- Worked with imaging group on Magnetic Resonance Elastography research
- Conducted MATLAB modeling to evaluate heart stiffness, presented findings at biomedical engineering poster session
Projects
- Built recommendation system using Gemma 270M and cosine similarity for personalized product matching
- Live at cigarstradamus.com
- Rewrote Facebook Prophet in Rust for better performance in production forecasting
- Running in production for mission-critical decision systems
- Available at ryanbieber.github.io/farseer
Skills
Agentic AI & LLMs:
LangGraph, LangSmith, LangChain, RAG, MCP Servers, Multi-agent systems, Conversational AI
Programming & Data:
Python, SQL, R, Rust, Statistical modeling, Time-series analysis
ML/AI:
NLP, Predictive analytics, Causal inference, Reinforcement learning, Generative AI
Cloud & Infrastructure:
GCP, AWS, Azure, Kubernetes (k8s/k9s), Helm, Docker, MLOps
Data Platforms:
Snowflake, Modern data warehouses, Analytics engineering
AI Governance:
Model validation, AI observability, Responsible AI deployment