Aurum Reserve

Inspiration

What it does

Inspiration

High-volume hospitality environments — from fine dining rooms like L'Étoile Noire to multi-channel kitchens handling drive-thru lanes, delivery dispatches, counters, and self-order kiosks — live and die by allocation accuracy. Most booking glitches happen when concurrent reservation systems write optimistically, resulting in costly double-bookings, chaotic service handoffs, and degraded guest trust.

We set out to build Aurum Reserve: an unapologetically reliable, zero-latency reservation command center that treats table and station allocations with the same strict transactional guarantees found in financial trading systems.

What it does

Aurum Reserve is a unified operational command center designed for multi-zone venues. It bridges front-of-house hospitality with high-throughput service hubs:

• Floor Command: Real-time visibility into active dining tables, drive-thru lanes, pickup counters, kiosks, and VIP lounges with instant status toggles and capacity metrics.
• Deterministic Allocation Engine: A best-fit seat optimization algorithm that maps guest party size $P$ to table capacity $C$ minimizing wasted capacity: $$\min (C - P) \quad \text{subject to} \quad C \ge P$$
• Atomic Concurrency Guarantee: Powered by SQLite configured with Write-Ahead Logging (WAL) and strict BEGIN IMMEDIATE transaction locks, ensuring absolute isolation and eliminating race conditions or double-bookings under peak traffic.
• Integrity Lab & Chaos Testing: Built-in stress test tools to simulate concurrent write spikes and verify lock integrity.
• Audit Ledger & Live SQL Console: Full operational traceability with an immutable booking ledger and an integrated query console for shift managers and operators.

How we built it

• Application Layer: Built in Python using Streamlit to create a dark-factory command cockpit with reactive state management and sub-second UI updates.
• Storage & Concurrency: Designed around an embedded SQLite engine utilizing WAL mode for high-concurrency read/write operations, enforcing transactional guarantees through explicit BEGIN IMMEDIATE locks.
• Data Modeling: Structured schema handling multi-zone partitioning (VIP, Counter, Drive-Thru, Delivery Hubs, Kiosks), timestamp serialization with timezone consistency, and guest contact indexing.

Challenges we faced

• Eliminating Race Conditions: In rapid multi-party reservation bursts, typical ORM-level checks suffer from time-of-check to time-of-use (TOCTOU) bugs. We tackled this at the database level by migrating logic into strict atomic transaction boundaries and benchmarking conflict handling.
• Balancing Versatility with Speed: Designing an interface that feels equally natural for managing an 8-seat executive lounge and routing high-frequency drive-thru or self-order kiosk capacity without clunky sub-menus.

Accomplishments that we're proud of

• Achieving a 100% collision-free record in high-concurrency simulation tests with zero double-allocations.
• Delivering an operator-first UI that unifies traditional dining reservations and automated service stations into a single command view.

What we learned

Deepening our understanding of database concurrency primitives, specifically the lock lifecycle of SQLite WAL mode under concurrent transactional write patterns.

What's next for Aurum Reserve

• Dynamic multi-venue load balancing.
• SMS/WhatsApp webhook dispatch for guest arrival check-ins and lane notifications.
• AI-driven floor turn-time prediction based on historical party sizes and order volumes. ## How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Aurum Reserve

Built With

• acid-transactions
• concurrency
• data-integrity
• operations
• python
• sqlite
• streamlit
• wal