Spencer Dobbs | Senior Analytics Engineer

The Challenge

When you're managing a portfolio of over 80,000 single-family rental properties, understanding the true state and potential of each asset becomes a monumental data challenge. Every property has its own story—its appliances, its maintenance history, its neighborhood dynamics, and its performance trajectory.

At Pretium, we needed a system that could capture all of this complexity and turn it into actionable intelligence for investment decisions.

What is Asset 360?

Asset 360 is an end-to-end asset modeling platform that I led the development of at Progress Residential. It's designed to:

**Track Property DNA** - Every characteristic of a property, from square footage to year built, from flooring type to HVAC systems

**Monitor Appliances** - Lifecycle tracking of all major appliances, predicting failures before they happen

**Model Operations** - Work orders, maintenance costs, and operational efficiency metrics

**Forecast Performance** - Predictive models that estimate future rent potential, maintenance costs, and overall ROI

The Technical Architecture

We built Asset 360 on a modern data stack:

**Snowflake** as our cloud data warehouse, enabling massive parallelization of complex queries

**Python** for data processing, ML pipelines, and custom analytics

**dbt** for transformation and documentation

**React** for the executive dashboard interface

The key insight was treating each property as an entity with a complete timeline of events, rather than just a collection of disconnected data points.

Impact

Today, Asset 360 powers critical decisions at the C-suite level:

Investment committee reviews rely on our performance forecasts

Disposition decisions are informed by our asset quality scores

Operational budgeting uses our predictive maintenance models

The platform processes over 10 million data points daily and maintains a 94% accuracy rate on 12-month performance forecasts.

Lessons Learned

Building Asset 360 taught me several key lessons about analytics engineering at scale:

**Data quality is everything** - Garbage in, garbage out applies at massive scale

**Business context matters more than technical elegance** - The best model is useless if stakeholders don't trust it

**Iterate with users** - Our best features came from sitting with portfolio managers and understanding their workflow

This project represents some of the most impactful work of my career, and I'm excited to bring these same methodologies to other Pretium portfolio companies.