Back to work
AI / MLPrototype

MissionCart

Goal-to-cart prototype that turns an occasion brief into a coordinated shopping list.

Type
AI / ML
Client
Hackathon prototype
Role
Product engineering · AI retrieval
Year
2026

Evidence / status

Prototype. Reported outcomes await verification.

Challenge

Shopping by intent, not keyword

Planning an event can require products from many unrelated categories. The prototype needed to translate one natural-language brief into a coherent cart within a fixed hackathon window.

Constraints

  • The team had 48 hours for a working prototype.
  • Product retrieval needed to understand occasion-level meaning.
  • The stack combined a React Native client with a Python service.
  • Budget, dietary preference, and availability needed to influence results.

Approach

Key decisions

  1. Structured occasion parsing

    The language model converts a brief into event type, guest count, preferences, and budget before product retrieval begins.

  2. Semantic catalogue retrieval

    FAISS retrieves product candidates by meaning rather than relying on exact keyword matches.

  3. Dedicated cart ranking

    A separate ranking stage balances relevance, availability, and price constraints before returning the cart.

  4. Scope fixed around a demonstrable journey

    The sprint prioritised one end-to-end occasion flow and a small set of supporting controls over breadth.

Evidence

What can be stated

Current status: Prototype. Unverified outcomes are labelled instead of presented as proof.

  • Existing project notes report six prototype features completed during a 48-hour sprint.

    Needs verification
  • Amazon HackOn 2026 finalist status requires owner verification before publication as proof.

    Needs verification

Technology stack

  • React Native
  • FastAPI
  • Groq
  • FAISS

Next project

Continue through the work archive.

Filio

Start a project

Let's build something great

Drop your details below. We'll have it open in Gmail, ready to send — no copy-pasting.