Codility Solutions
Proof of Work

Case Study Pack

AI and SaaS Products Shipped to Production

Real problems solved. Real systems built. Real outcomes delivered - by the engineering team at Codility Solutions.

4 Case Studies
13 Projects Shipped
6+ Industries
AI Voice / Mental Wellness

JustListenly

AI Voice Companion Platform - justlistenly.com

The Problem

Voice AI products typically suffer from 2-4 second response latency that destroys the sense of presence in empathetic conversations. Raw LLM output creates repetitive, robotic phrasing unsuitable for emotional support interactions. The client needed a voice companion that felt human - with payment automation and operational tooling built in from day one.

What We Built

  • 4 distinct AI personas with unique voices via ElevenLabs (Grandpa Jim, Grandma Grace, Mom Rose, Dad Mike)
  • Sub-2-second response latency using first-sentence-first TTS and static prompt pre-caching
  • Repetition detection and problem phrase filtering for natural conversation quality
  • Crisis detection routing callers to the 988 mental health line automatically
  • Emotional theme extraction for cross-call memory and relationship continuity
  • Systeme.io to Stripe payment pipeline with heuristic matching and automated billing
  • Session-based subscription model with explicit consent for overage charges
  • 5 SMS automation workflows for engagement and retention
  • Django admin panel for non-technical operations staff

Tech Stack

Node.js / Express Twilio ElevenLabs OpenAI PostgreSQL Redis Stripe Django Docker PM2 / Nginx

Outcomes

  • Perceived latency reduced to under 2 seconds
  • Higher user retention with preferred personas
  • 67+ unit and integration tests across core services
  • Fully automated, auditable payment flow from sale to active subscription
  • Non-technical staff can operate the platform independently
Codility Solutions - Case Study Pack
AI Voice / Qualitative Research

Impact Intelligence

AI Interview Operating System - impactintel.com

The Problem

Manual interview operations created significant bottlenecks across scheduling coordination, transcription workflows, no-show handling, and the slow conversion of raw interviews into actionable insights. Traditional research methods could not scale for high-volume, multilingual stakeholder conversations across multiple regions. The team needed a system that could run interviews without human intervention.

What We Built

  • On-demand AI voice interview system with secure participant access via unique expiring links
  • Multilingual support across major language variants for global stakeholder reach
  • Automated call state tracking and reconciliation - no manual follow-up needed
  • Batch transcript export for rapid insight generation
  • Role-based access control and per-application quota management
  • Per-application isolated Cloud Run LLM endpoints for security and scale
  • Concurrency-aware call admission checks to prevent overload
  • Full Django admin panel for non-technical research operators

Tech Stack

Rails 7.1 React Retell AI FastAPI Python OpenAI Google Cloud Run PostgreSQL ActiveAdmin

Outcomes

  • Eliminated scheduling bottlenecks entirely
  • Multilingual AI interviews running at scale across regions
  • Automated transcript pipeline from call to structured data
  • Research operators run sessions without engineering support
  • Direct data pipeline from interviews to strategic decision-making
Codility Solutions - Case Study Pack
B2B SaaS / Corporate Food Delivery

Chowmill

Multi-Sided Corporate Food Delivery Platform - chowmill.com

The Problem

Chowmill needed to manage three entirely different stakeholder workflows simultaneously - employees ordering within budget constraints, restaurants managing orders and logistics, and administrators orchestrating multi-location operations. Manual meal scheduling for large companies across multiple offices was operationally unsustainable. A single misstep across any stakeholder layer cascaded into failures across all three.

What We Built

  • 3 independent React portals sharing a single Rails API with role-scoped endpoints
  • RecurringScheduler engine automating daily meal event generation without manual intervention
  • Onfleet and Senpex delivery integration with chained pickup-dropoff task graphs
  • Multi-channel notification system (FCM push, Twilio SMS, email) with full logging
  • Stripe billing and QuickBooks sync handling dual billing (company invoicing + restaurant payouts)
  • Elasticsearch-powered menu search with event-driven reindexing
  • Incremental reporting covering nutrition, satisfaction, vendor performance, and budget tracking
  • 630+ database migrations supporting domain maturity over time

Tech Stack

Rails 7.0 React / TypeScript PostgreSQL Sidekiq / Redis Elasticsearch Stripe QuickBooks API Twilio AWS S3 Docker

Outcomes

  • Production platform handling complex multi-sided operations with no manual escalations
  • Restaurants achieve full self-serve order management and billing access
  • Companies automate recurring meal scheduling across locations and employee segments
  • Full audit trail and compliance tracking across all critical domain models
Codility Solutions - Case Study Pack
AI / Hospitality Procurement

Orders Made Simple

AI-Powered Food Service Procurement Platform - ordersmadesimple.com

The Problem

Restaurant groups managing multiple venues faced fragmented ordering across disconnected portals, phone lines, and email chains. With 60+ supplier relationships to manage simultaneously, missed and duplicate orders were common. Suppliers were manually entering orders from multiple channels - consuming roughly 400 staff-hours per month in avoidable data entry. There was no unified view of cost, inventory, or supplier performance.

What We Built

  • Mobile app (React Native / Expo) for kitchen-floor ordering, invoice capture, recipe costing, and spending tracking
  • Web dashboard (React + Ant Design) with order history, gross profit calculator, Xero sync, and receipting workflows
  • Supplier portal with AI co-pilot for automated email order processing using GPT-4 and Tesseract OCR
  • 80+ Lambda functions handling full order lifecycle, catalogue management, and billing
  • Integrations service routing orders to 13+ suppliers in their required formats (XML / CSV via SFTP / S3 / Email)
  • Real-time updates via Socket.IO and Redis pub/sub across all platform components
  • Multi-region AWS Cognito authentication with zero-disruption auto-migration

Tech Stack

React Native / Expo React / Ant Design Node.js AWS Lambda MongoDB GPT-4 Tesseract OCR Socket.IO Redis Xero API Stripe AWS Cognito

Outcomes

  • Single unified interface replacing 6 separate supplier portals
  • 400 staff-hours per month eliminated via AI order capture
  • 13+ direct supplier integrations with automatic format routing
  • Real-time gross profit tracking and spend visibility
  • 80+ Lambda functions deployed and running in production
Codility Solutions - Case Study Pack