Abstract AI Pipeline Studio
Client project: Abstract Security (Crest Data Client Engagement)
Technical Deep Dive
Owned end-to-end frontend architecture for a multi-tenant pipeline platform with visual DSL workflows, contract-first API integration, and retry-safe configuration delivery.
Client Context
Security teams needed a no-code system to configure high-volume telemetry pipelines safely across AWS, Azure, CrowdStrike, Tenable, and ServiceNow integrations.
Execution
Designed a React + TypeScript visual DSL that emitted deterministic, versioned JSON AST definitions, enforced schema validation and RBAC-aware editing, and standardized service access with OpenAPI-generated types plus normalized Axios error handling.
Outcome
Delivered safer pipeline rollouts with idempotent saves, version-aware concurrency checks, and guarded NLP query generation while lowering client-side integration errors by about 60%.
“Owned and built the entire React + TanStack Query frontend from scratch, including an NLP-assisted node graph that translated plain-English conditions into SQL-like logic and AI-suggested source/destination/enrichment mappings, while integrating 50+ FastAPI services and reducing client-side errors by ~60%.”
Core Stack
Metrics
services
50+
error_drop
60%
tenant_scale
Hundreds of pipelines/org
save_model
Idempotent + version-aware