Problem

Equity analysts at a boutique firm tracked disparate datasets (CSV, vendor APIs, custom notes) and stitched them manually in Excel. Leadership wanted a single workspace with smart tagging and alerting to surface exposure risks within seconds.

Approach

• Designed a Go service graph with domain-driven modules and OpenAPI-first contracts.
• Built ingestion workers that sync holdings into CockroachDB, denormalizing for OLAP-style queries.
• Implemented vector tagging (OpenAI embeddings cached via Redis) to cluster similar positions and annotate market narratives.
• Crafted a responsive Vue 3 SPA with deck.gl visualizations and an AI assistant for natural-language queries.
• Observability: Honeycomb tracing, Loki logs, Terraform provisioning for GKE.

Results

• Aggregations under 1.2 seconds for 50k holdings across four asset classes.
• Analysts reported 22% faster research sprints and fewer missed compliance triggers.
• Delivered multi-tenant RBAC and per-client branding delivered via feature flags.

Links

• Source repository
• Live demo