BACK TO SELECTED WORK
CASE STUDY 04 / 09Platform / Product
Secure internal AI operations workspace

SankhyaAI

A production-style AI support console with authenticated chat, document ingestion, admin observability, SSE streaming, RBAC, audit logs, and persistent state.

Next.jsFastAPIPostgreSQLRedisGeminiSSERBAC
SankhyaAI project previewPROJECT / SPLIT
01 / THE WORK

Behind the build.

SankhyaAI is more than a chat UI. It combines a Next.js workspace, FastAPI backend, PostgreSQL persistence, Gemini responses, role-based access control, audit logging, and an admin dashboard for document ingestion and operational visibility. The product is structured like an internal support copilot that can be extended into a retrieval-enabled operations assistant.

02 / ENGINEERING DECISIONS

What makes it work.

01 / CAPABILITYAuthenticated user workspace with persisted conversations
02 / CAPABILITYAdmin console for document ingestion, users, metrics, and system health
03 / CAPABILITYSession-backed auth, RBAC, and audit logs for sensitive actions
04 / CAPABILITYGemini-powered streaming responses with optional Redis-backed workflows
03 / INTERACTIVE WALKTHROUGH

From input to outcome.

STAGE 1 OF 5

Users sign into the AI workspace with seeded local identities.

04 / ARCHITECTURE

Under the hood.

Enterprise Support Copilot Architecture & Audit Loop

Security Diagram 4.1
1. Auth Workspace

Next.js UI + RBAC

Session authentication and role-based permissions.

2. API Gateway

FastAPI + Redis

SSE streaming orchestration & document ingestion.

3. LLM Engine

Gemini Model

Contextual answer generation & document Q&A.

4. Audit & Storage

PostgreSQL Persistence

Audit logs, conversations, token accounting.

05 / RESULTS & USE CASES

Built to matter.

OUTCOME 01A secure base for internal AI copilots
OUTCOME 02Clear separation between user and admin operations
OUTCOME 03Production-minded persistence, observability, and access control
Support teamsInternal operationsDocument Q&AAdmin AI workspaces