Quadrant Technologies · Internship · 2026
AI HR Copilot
An agentic HR assistant that screens résumés, routes tickets and tracks regulatory compliance, with a human approving every consequential action.
- Python
- FastAPI
- Next.js
- Azure Cosmos DB
- MCP
- OpenHands
- Terraform
- teams, selected for production rollout
- 1st of 9
- tool capabilities
- 149
- MCP integrations
- 11
- employees served
- 400+
The problem
Résumé screening, HR ticketing and regulatory compliance tracking needed automating for an organisation of 400+ employees, without letting an AI agent act on anything a person had not approved.
My role
I led backend engineering on a team of interns.
What I built
- A Python FastAPI backend connected to a Next.js frontend and Azure Cosmos DB, with strict PII redaction on résumé, ticket and compliance data.
- A tool-calling architecture exposing 149 capabilities across 11 MCP integrations.
- Human-in-the-loop guardrails so the agent proposes and a person approves.
- Data flow and availability work using OpenHands, sized for 400+ employees.
The hard part
Giving the agent 149 tools while keeping it safe and cheap: every consequential action waits for human approval, and monthly cloud cost had to stay under budget.
Why it matters
It placed 1st among 9 teams and was selected for live production rollout, so the design had to hold up for real employees rather than a demo.
How it was measured
Judged against 8 other intern teams and chosen for rollout, with monthly cloud overhead held under budget.
Architecture
- 1Next.js appChat, intake and checklist views
- 2FastAPIAgent orchestration and auth
- 3MCP tools149 capabilities, 11 integrations
- 4Human approvalNothing consequential runs unapproved
- 5Cosmos DBCandidates, tickets, compliance state