Custom AI Solution for Project Mngmt
DoorDash asked me to re-imagine project management for its 50-project Marketing Tech org—without disrupting existing tools or workflows.
Over one intense weekend I designed a serverless, AI-driven RAG pipeline that:
- Auto-harvests updates from Jira, Slack, Gmail & Docs
- Classifies, embeds, and stores them in BigQuery with near-real-time vector search
- Generates persona-specific digests (Exec, PM, Engineer) via Gemini 2.5 Pro
- Audits itself with LLM-based relevance & fact-check judges to guarantee quality
Impact model: ~10 engineer/PM hours reclaimed per team each week → 407–1,087% ROI at launch-scale, with <$300 / month cloud spend.
I relied on 85% low-code GCP services (Application Integration, Workflows YAML, Vertex AI Agent Builder) plus a slim Cloud Run worker for embeddings—proving that enterprise-grade AI automation doesn’t have to wait for a full dev sprint.
Dive into the architecture below to see how low-code, secure design can unlock velocity and clarity for cross-functional teams.
Modern ops excellence isn’t just about adding more meetings or dashboards—it’s about turning your existing data exhaust into real-time intelligence. By fusing no-code integration, vector search, and LLM “quality gates,” this design shows how any marketing org can reclaim hours, surface blockers instantly, and deliver stakeholder-ready insights—all for the cost of a few cloud instances. AI isn’t the future of project management; it’s the upgrade your tech stack can adopt today.