Personal Picker
Personalized logistics and shopping assistant app.
What we built
and why.
Personal Picker is a personalized logistics and shopping assistant that learns from user behavior to recommend the right products, delivered by the right route, at the right moment. Built on Flutter + Node with a recommendation engine that updates with every interaction.
The problem
to solve.
Context
Logistics · Personalized Commerce : The client saw shoppers drowning in choice and logistics companies drowning in inefficient routes. They wanted one app where personalization solves both problems: customer preference and delivery reality converging in real time.
Core Problem
Most shopping apps recommend generically and route deliveries statically. The opportunity was to personalize the entire loop: what you see, what you buy, and how it gets to you.
How we
built it.
An 8-week build. Flutter for fast cross-platform iteration, a Node.js backend with a streaming recommendation service, and a logistics layer that treats delivery as a first-class feature.
Data Model
Designed the personalization schema: user preferences, purchase history, delivery signals, in one unified graph.
Engine & UX
Built the recommendation service with live re-ranking and a shopping UI that reflects changes within seconds.
Logistics
Integrated delivery routing with customer preference (time window, location, preferred carrier).
Launch
Staged launch in one region, measured, expanded.
What got
shipped.
Flutter app consuming a Node/Express API backed by Postgres (source of truth) and Redis (recommendation cache). A dedicated recommendation microservice streams updates over WebSockets so the feed stays live without refresh.
Key Innovations
- Live-streaming recommendations that re-rank as the user taps through categories
- Delivery-aware product surfacing (nearby + available = higher rank)
- Preference learning from implicit signals (dwell time, category revisits)
- Driver-side companion app sharing the same backend
Obstacles Overcome
- Keeping recommendation latency under 300ms while re-ranking on every event
- Handling delivery partner dropouts gracefully mid-order
- Training the initial recommendation model with sparse cold-start data
What it
does.
5 core capabilities that define the product. Each engineered with a senior team, tested against real usage, and shipped to production.
Interest-Based Matching
Recommendation engine suggesting products based on demographic data.
Occasion Tracking
Integrated calendar and alerts for birthdays and holidays.
Retail Integration
Seamless connectivity with major e-commerce platforms like Amazon.
Relationship Mapping
Gifting logic that adjusts results based on recipient relationship.
Budget Optimization
Dynamic filtering to find high-value gifts within constraints.
The product,
end to end.
11screens from the shipped build. Every flow, every state. These aren’t renders, they’re production.
The impact,
measured.
Turned a generic shopping app into a personalized assistant that respects time and preference, increasing engagement on the consumer side and utilization on the logistics side from the same codebase.
Built with.
Personalization works when it's fast, transparent, and wired into the whole journey, not just the homepage. Personal Picker delivered that end to end.
Got a project that
needs this kind of build?
Tell us the problem. We’ll tell you if it’s a 2-week sprint or a 2-month platform, honestly, in the first call.
