Food Analytics

Data analytics platform helps our customers around the globe to reduce costs and improve resurce planning on daily bases.

Services

Product Management
Project Management
System Architecture
UI/UX Design
Web Development
Data Analytics

Tech and tools

Java
Spring Boot
React
Javascript
Typescript
GraphQL
GRPC
Druid
Mongo DB
AWS
Docker
Kubernetes

Project Information

Food Analytics platform

Our customer needed a platform to better analyze its millions of annual visitors, provide better offers, and streamline internal procedures to save money and improve quality. Visual KPIs based on computer vision, such as portion size, right dish arrangement, plate cleanliness, and many more, were added in addition to detailed sales KPIs, thanks to the daily analysis of millions of photographs. In addition to a user-friendly frontend, the complete backend was built using modern technologies to process enormous volumes of data in the cloud effectively.

The Solution

Food analytics is a fully customized data analytics platform for in-cloud data analysis. Helps businesses and companies to have a better overview of restaurants from a bigger picture perspective. It helps them to plan the restaurant’s budget and organize employees.

Data is prepared and aggregated from all restaurants on a daily basis, and reports can be generated from one restaurant to the whole organization.

Main features

The system introduces real-time monitoring of food servings, with different user roles having the ability to monitor and mark correct/incorrect servings, as well as provide extra product information and comments. When the teams in charge are working on dataset preparation, these corrections can be useful.

The system provides statistics and analytics insigts on everything from the most popular food and drink combos to the most popular side dishes on the plate. Stakeholders can forecast how much food and ingredients will be required on specific periods of the day, week or requirements for even months ahead.

pTag

A system that increases the efficiency of OPS teams by automating 80% of tasks.

Services

Product Management
Project Management
System Architecture
UI/UX Design
Software Development

Tech and tools

Java
Spring Boot
Python
React
Javascript
Typescript
GraphQL
GRPC
REST
Druid
Mongo DB
Postgres
Redis
AWS
Docker
Kubernetes

Project Information

Dataset Management

pTag as a service was created to assist data operations teams in resolving issues and decreasing dataset preparation time while improving data quality.

The service works with a variety of dataset types to prepare data for various machine learning algorithms. The classification dataset is created on crops generated from bounding boxes annotated on full frames, whereas the detection dataset is built on complete frames.

Main features

pTag can integrate with third-party computer vision annotation tools to create bounding boxes for ML training.

The dataset service includes a cropping function that is used for crop generation based on bounding boxes. Clustering is the following step once cropping is completed. Clustering is a feature that groups crops based on their similarity. The amount of time it takes to prepare a dataset is dramatically decreased using this functionality.

Our team created a dataset cleanup function that detects and removes duplicate and blurry photos, resulting in a high-quality dataset. Additionally, dataset versioning is provided to aid in the tracking of changes to datasets as well as the ability to merge two or more datasets if necessary.

ML Platform

We designed and developed seamles solution for Machine Learning process, model management and deployment.

Services

Product Management
Project Management
System Architecture
UI/UX Design
Prototyping
Software Development

Tech and tools

Java
Spring Boot
React
GraphQL
GRPC
AI
ML
AWS
Docker
Kubernetes

Project Information

Machine Learning as service

A simple and feature-rich platform was required for one of Europe\'s fastest-growing AI firms. The goal was to create a user-friendly solution that would allow even non-tech folks to quickly train models and modify parameters.

Biggest challenge was to create an end-to-end platform capable of managing the entire machine learning lifecycle, including dataset preparation, experimentation with various models and algorithms, real-time monitoring of the entire training process, testing and reproducing an outcome, and finally, automatically or manually deploying the model to test devices based on defined rules.

Sweet spots

The technology may detect a potential problem during training and provide real-time metrics as well as options to halt or stop training, test specific weights, and only continue training if necessary, saving time and resources.

The system incorporates a couple of layers of validation and a comprehensive testing framework built as a quality-gate before final deployment.

For quick model deployments and version management, the platform enabled connections with in-house and third-party systems. It covers multi-level client and user administration, inventory and product management, as well as integrated solutions such as Auth0 or other bespoke authentication providers.

Remedim

Device Management for IT admins to monitor and maintain a variety of devices in a secure and timely manner.

Services

System Architecture
Edge Device development
Software Development

Tech and tools

Java
Spring Boot
Python
React
Javascript
Typescript
GraphQL
GRPC
Postgres
Redis
AWS
Docker
Kubernetes

Project Information

Edge Device Management System

Our customer, which has over one million edge devices installed worldwide, wanted a real-time operations platform. The requirements included remote access for device configuration, as well as maintenance and monitoring of all its devices around the world. Major focus and requirement was security. Our team designed the system so that all devices are securely connected to internal service via RPC bi-directional protocol and communication is enabled from both sides, from client to server and vice versa.

The device automatically connects to the internal service after it is installed and linked to the internet. The custom-developed web platform allows for maintenance and monitoring. Custom deployments and device updates can be carried out on a single device or numerous devices at once.

Main features

Dynamic configurations for each device, with a clever scheduling system, configurations may be simply altered and updated to one or more devices at the same time..

Device hardware monitoring and alarm with predefined thresholds.

Real-time device camera monitoring with a scanning region based on a predetermined ROI. This function is extremely useful for IT administrators who want to double-check that camera locations are right.

IT admins can trace and simply restore prior configurations if the new one isn\'t operating as planned thanks to device configuration versioning.

AI Segmentation

Data segmentation platform that automatically spots outcome-focused customer segments, and enables adaptable product engagement in minutes.

Services

UI/UX Design
System Architecture
Software Development
AI Research & Development

Tech and tools

Python
React
Javascript
Typescript
Tensorflow
Spark
Clustering
Supervised and unsupervised learning
MongoDB
PostgreSQL

Project Information

Data segmentation platform

Our customer needed a platform that acts as an AI co-pilot for segmentation by automatically recommending key user segments, understanding effectiveness of each segment and highlighting their impact on key business KPIs. The Platform automatically syncs dynamic segments to application code through first-class SDKs, CRM platforms (like Salesforce), email marketing platforms (like Marketo & Iterable) and product engagement tools (like Pendo).

The Solution

AI segmentation helps companies solve their most common obstacles for the success of low-touch customers by tracking every one of their customers' objectives and it's unique data centralisation features that bring product, CRM and warehouse data together in minutes.

The Platform uses the power of the AI engine to automatically spot key customer segments while enabling companies to identify account risks early on through platform's robust segment analytics and automated signals.

Main features

The Solution introduces automated goals-based dynamic segmentation with AI algorithms that are fine tuned to their customers’ business to automatically detect key customer segments including churn-risk accounts, expansion accounts and many more.

No-code platform helps users create in-product and out-of-product experiences (including webinar registrations, renewal warnings, customized product journeys and onboarding), all tailored to adaptable segments which are constantly being monitored, analysed and further improved on their own.