Enterprise Workflow Assistant Platform
A multi-tenant platform for managing internal knowledge, business tools, and model settings by workspace.
Challenge
Internal knowledge search and routine operations were split across teams and tools, making it hard for users to find the right information or workflow. The platform needed enterprise-grade authentication, tenant isolation, monitoring, and configurable assistants for each business use case.
Solution
Structured the backend around clean architecture and built an internal execution environment on GCP. Separated assistant execution, knowledge retrieval, code sandboxing, model switching, and workspace-level access control.
Result
Delivered a full-stack enterprise platform covering frontend, backend, and infrastructure, ready to operate assistants by department.
Team
1 member, 5 months
Design, implementation, infrastructure, operations
Role
Handled product design, implementation, infrastructure, and operations design.
Led frontend (Next.js), backend (FastAPI), and infrastructure (Terraform/GCP) end-to-end.
Tech Stack
Key Features
SSE streaming chat: response generation, tool execution, and errors displayed as separate events
Multi-model support: Switch between OpenAI / Anthropic / Gemini through a unified interface
Assistant settings screen: manage system prompts, tools, and models from the admin UI
Knowledge base: vectorize documents and retrieve related context during responses
Code sandbox with RestrictedPython: Secure user code execution environment
Multi-tenancy: Structurally prevent cross-tenant access with workspace_id filtering across all tables
Technical Highlights
Clean Architecture
Unified dependency direction from domain → application → infrastructure → api, enabling LLM provider switching with only infrastructure layer changes.
Enterprise Security
Restricted backend to internal traffic only, implemented dual authentication with OIDC + user tokens, DB isolation via VPC, and secret management via Secret Manager.
Cost Optimization
Reduced development environment costs to 2-5% of production through Cloud Run scale-to-zero and environment-specific resource isolation. All infrastructure managed as code with Terraform.