Boomerang AI

Professional case study covering B2B SaaS engineering, data enrichment, and relationship intelligence systems.

Data Ingestion → Entity Resolution → Enrichment → API Delivery

01 — Overview

During my time at Boomerang AI, I focused on building and scaling backend systems that drive relationship intelligence. This case study highlights the general engineering problems solved in a high-throughput B2B environment.

02 — Problem

B2B platforms require processing massive amounts of unstructured external data (job changes, relationship signals) and mapping them accurately to internal CRM structures. The system must be incredibly fast, highly reliable, and cost-efficient.

03 — Why it was technically difficult

The primary challenge was entity resolution and data enrichment at scale. When pulling from disparate external APIs and data providers, mapping fuzzy signals (like a company name variation or a job title change) to a canonical database without introducing false positives requires sophisticated architecture.

04 — Architecture

A distributed pipeline architecture designed to ingest webhooks and polling data, process it asynchronously, resolve entities, and serve the enriched data via high-performance APIs.

05 — Key engineering decisions

• Distributed Pipelines: Utilized asynchronous workers to decouple data ingestion from processing, allowing the system to absorb massive traffic spikes.
• Cost Optimization: Implemented aggressive caching and deduplication layers before hitting expensive third-party enrichment APIs.
• API Performance: Optimized database queries and indexed heavily used paths to ensure sub-100ms response times for customer-facing endpoints.

06 — Product decisions

• Accuracy over Speed: In B2B data, a wrong signal is worse than a delayed signal. We optimized our entity resolution for high precision, explicitly flagging low-confidence matches for review.

07 — Outcome

Contributed to a highly scalable infrastructure capable of processing millions of data points, directly enabling core features of the Boomerang AI platform.

08 — What I learned

Working at Boomerang AI reinforced the importance of API performance and the hidden costs of data pipelines. I learned to design systems defensively, assuming that external data will always be messier than expected.