DesignPro
Idea to design, guided by AI
Using AI and LLMs to understand design requirements and feedbacks to help designers manage their work
The Challenge
Help DesignPro become a one stop shop for designers by helping the designers understand design requirements, give them feedback on designs and help them manage their workflows.
The Outcome
We explored prompt engineering, chaining and finetuning LLMs like LlaMA-v2 and OpenAI GPT-4 to prototype and research different product ideas. We developed multiple prototypes and investor demos integrating LLMs with Figma APIs developing custom plugins within a single quarter.
Our Impact
Our engineering efficiency and AI expertise helped the team develop a quick prototyping and research pipeline allowing the startup to validate their product ideas and theories with real customers.
Nauyan Rashid
Machine Learning Engineer
$9,500 / month
Generative AI
LLMs
Figma API
Python
Machine Learning
ALL3D Scanning Application
A One Stop Shop for 3D Content
Our 3D reconstruction and scanning solution helps jump start 3D model creation process
The Challenge
Help 3D artists create the models more easily and efficiently by harnessing the power of 3D scanning and computer vision.
The Outcome
We implemented a 3D scanning and deep learning based 3D reconstruction pipeline to automate the creation of 3D models and spaces. We researched deep learning models for Shape Analysis and Completion (NeRFs, 3D Reinforcement Learning and Differentiable Rendering) and Material and Texture Synthesis (Light estimation, pattern recognition and material classification)
Our Impact
We allowed normal users to create 3D assets using a simple scanning application without requiring knowledge of 3D and helped designers create 3D assets in a fast and efficient manner.
Muhammad Ali
Lead Computer Vision Engineer
$12,000 / month
Computer Vision
CoreML
3D Reconstruction
PyTorch
Tintash
Generative Game Art
A custom Trained Stable Diffusion Model to generate Game Art Assets like Spirtes, Icons etc.
The Challenge
Creating art sets for games specially large number of icons and sprites that should adhere to the same art style takes lots of time and a large art team.
The Outcome
Our domain experts finetuned a specialized Stable Diffusion model using LoRA and ControlNet that was capable of generating thousands of Icons, Spirtes and assets for game development greatly increasing the productivity of game artists.
Our Impact
The client was able to include AI generation as a force multiplier in their Art pipeline and has been using it to generate whole sheets of ready to use Icons and Spirtes in any art style needed.
​
​
AI Generated Game Art and Variantes
Shahryar Saqib
Games & AI Engineer
$10,500 / month
Game Development Generalist
Generative AI
Unity
Programming
AI for Games
Nitrocam
Video Analytics as a Service
Our visual analytics platform empowers any camera with the power of AI
The Challenge
Develop a computer vision analytics solution to automate video surveillance and activity monitoring at scale without purpose built expensive cameras.
The Outcome
Our video surveillance system as a service (VSaaS) solution was designed to be cloud agnostic and scalable. We were able to achieve a video latency of 300 ms allowing the system to work in real time. We implemented state-of-the-art solutions for occlusion detection, object and person tracking and recognition, activity recognition and monitoring, and geospatial mapping of events. We also allowed users to create customisable rules for video and event monitoring.
Our Impact
We enabled Nitrocam to provide its users with low-cost cameras that still offer the features of any state-of-the-art security camera. The modular solutions we developed allowed Nitrocam to reduce their costs by selecting specific services for their customers and apply monitoring and video surveillance to any pre-recorded videos. This has had a significant impact on Nitrocam’s customer acquisition, as they are now able to offer a more affordable and customizable video surveillance system that meets the needs of a wider range of customers.
Muhammad Ali
Lead Computer Vision Engineer
$12,000 / month
Computer Vision
OpenCV
Video Analytics
PyTorch
World’s Largest Furniture Retailer
Sales forecasting and price optimization
Sales forecasting and price optimization solution to help guide logistics & pricing decision making
The Challenge
Help optimize supply chain logistics and operations of the sales team by improving procurement and pricing strategies.
The Outcome
We implemented a demand forecasting solution that consisted of thousands of in-database time series models running in parallel to update the predicted sales numbers. Product affinity and cross price elasticity were also calculated based on historical sales data in order to serve as input to price optimization models.
Our Impact
We enabled business decision makers by providing them with sales forecast numbers for over 10,000 SKU’s and their optimal pricing structure so that resources could be allocated in a more efficient manner.
Sales Forecasting
Price Optimization
Asif Jamshed
Lead Data Scientist
$11,000 / month