Benjamin Fuller

Problem

A real-estate investment operation needed a platform to evaluate properties and portfolios and act on buy/sell/hold decisions, and wanted it built fast, using AI agents to do much of the engineering with humans in the loop.

Approach

Building the portfolio evaluation and trading platform (React/TypeScript/Node on a cloud data pipeline) while pioneering an autonomous engineering workflow: AI agents doing ETL, analysis, and implementation behind human-in-the-loop review gates, so both the product and the way it's built are AI-native.

Technologies

Problem

Urban business- and general-improvement districts run on manual administration (assessments, services, reporting, and stakeholder communication). There's a real opportunity to run one far more efficiently with automation and AI.

Approach

Designing and building an AI-powered operating model for a neighborhood improvement district, automating administration, operations, and stakeholder workflows end to end.

Problem

A professional care practice was losing time to phone reception, manual scheduling, email triage, and administrative overhead that pulled practitioners away from their clients.

Approach

Built a voice-automated receptionist and a set of AI tools for the practitioners: automating intake and scheduling calls, email and task management, and routine admin, with human oversight where it mattered.

Problem

A software company wanted an AI assistant inside their product so users could ask questions and get accurate answers grounded in the product's own data, without hallucination.

Approach

Integrated a conversational assistant with retrieval-augmented generation over the product's domain data, with grounding, guardrails, and evaluation so answers stayed accurate and on-topic.

Problem

Real-estate investors and operators decide whether to hold, sell, refinance, or renovate each property using data scattered across property-management and accounting systems, so the analysis is slow, manual, and inconsistent from asset to asset.

Approach

Built a pipeline that pulls from AppFolio and QuickBooks into BigQuery, computes per-asset financials, and scores each property's best next move, surfaced through a Looker layer the team can act on.

Technologies

Problem

A core billing process was manual and multi-step, taking about three days of human effort each cycle and leaving room for error.

Approach

Built a multi-agent system (RAG and MCP integrations) that runs the billing workflow end to end, with human-in-the-loop checks at the points that matter.

Problem

Teams adopting AI tools move fast but without shared guardrails, risking quality, security, and inconsistent results as adoption scales.

Approach

Established reference patterns, guardrails, and enablement programs aligned to NIST AI RMF: evaluation in the workflow, least-privilege access for agents, and human-in-the-loop review.

Technologies

Problem

Every parcel leaving a warehouse was a missed marketing opportunity: e-commerce brands had no way to turn the physical shipping label into an owned, addressable channel.

Approach

Co-founded the venture and led engineering as CTO, building the full stack (a Shopify app for brand onboarding and campaigns, a Chrome extension, a ShipStation desktop integration for 3PL hubs, and a NextJS/React advertiser dashboard) on patented QR-driven label technology, and drove the live pilot integrations with ShipStation and USPS.

Technologies

Problem

A prospecting platform is only as good as its search: users have to find and rank the right B2B contacts out of hundreds of millions of records, instantly.

Approach

Built the production ranking and lead-scoring engine from the first commit on AWS, tuned for sub-second retrieval across the full dataset so prospect search felt instant.

Technologies

Problem

After the acquisition, a live product with paying customers had to be merged into the acquirer's multi-product platform without downtime or disruption.

Approach

Led the integration incrementally so customers kept working throughout, coordinating a cross-continental team (US, India, Europe) through their leaders.

Technologies

Problem

High-volume mobile location data arrives continuously and has to be ingested, processed, and delivered at scale, reliably and cost-effectively.

Approach

Built an event-driven streaming platform on AWS (Kafka, Spark on EMR, ECS) designed for high throughput, reliability, and cost control, under SOC 2.

Technologies

Problem

Location data at scale is noisy: fraud, bot traffic, and low-quality signals degrade the product if they reach customers.

Approach

Built proprietary data-quality, fraud-detection, and bot-detection algorithms directly into the streaming pipeline, catching bad data upstream rather than after delivery.

Technologies

Problem

Onboarding a new data client took 3+ months of manual, project-managed work, capping growth and tying up the team.

Approach

Shipped a self-serve portal that let clients configure, onboard, and pull data themselves, removing the manual project plan from the critical path.

Technologies

Problem

Regulated financial and mortgage data platforms across four business units ran on aging infrastructure that was costly and hard to scale.

Approach

Led the modernization and re-platforming onto AWS across the business units, keeping delivery running throughout, with a cloud-first template that scaled org-wide.

Technologies

Problem

Pricing single-family rentals accurately at scale requires turning raw rental-market data into reliable, defensible valuations.

Approach

Led development of the Rental Automated Valuation Model at RentRange, where the model is the product; the method was novel enough to be patented.

Problem

Institutional investors buying single-family homes need to source, value, and underwrite properties against a defined buy-box far faster and more consistently than manual analysis allows.

Approach

Led Investability, an automated buy-box, property-matching, deal-scoring, and underwriting platform that sourced, valued, and acquired single-family homes end to end.

Technologies

Problem

A NASDAQ company growing by acquisition has to fold acquired companies' technology and teams into the enterprise platform without breaking either side.

Approach

Led the technical side of roughly seven M&A integrations: due diligence, integration, and separation, bringing acquired systems and distributed teams onto the platform.

Technologies