Projects

Here is a sample of projects we have executed in each of our three areas of services: Software Technology Strategy Analysis, Data Acquisition Strategy Consulting, and Software System Design and Implementation. We will be happy to discuss your specific needs and benefits of using advanced software, artificial intelligence, and data analytics technology in your products and company operations. 

Software Technology Strategy Analysis

Using Artificial Intelligence in Procurement Software Systems. After performing an ind-dept analysis of the customer product line compared with other products available in the market we advised on a strategy and a multi-year product development road map to use AI, Machine Learning, and Natural Language Processing technology to improve the customer products position in the marketplace.

Quantum Computing Software. We analyzed our Japanese client’s quantum computing technology within the quantum computing space (both available products and research efforts), and then we recommended specific high-value problems that can best be reduced to forms that can be efficiently solved on the client’s quantum computing architecture.

Enhancing Datacenter Security Software with AI Technology. After a thorough analysis of our Japanese client’s data center security product in the marketplace, we suggested specific projects to enhance the effectiveness of the product using semantic analysis and machine learning technology.

Data Acquisition Strategy Consulting:

Cryptocurrency Arbitrage Using Artificial Intelligence:  From a comprehensive analysis of our client’s trading goals and methodology, we recommended a complete set of data to be acquired and stored by our client a couple of years in advance of them bringing online new state-of-the-art trading systems which we will develop using machine learning semantic analysis, and data analytic technology.

Retina Eye Treatment Using Machine Learning Technology:  After a thorough analysis of our client’s eye surgery device, we recommended a comprehensive set of data to be collected in order to enable our client to use in future versions their product advanced deep learning and computer vision technology that will make their product more competitive and propel it past its competitors’ position in the market.

Software System Design and Implementation

Text-Based Question Generation. A consortium of engineering companies was looking to augment the set of teaching tools for engineering domains. While question answering is an effective way of improving reading comprehension, tutors have limited time for asking individualized questions and providing personalized feedback. Our team built a software system that generates relevant questions of varying complexity for a given text using Large Language Models (LLM), Natural Language Generation (NLG) methods, and validation by a human tutor.

Car damage predictions. Our customer, an insurance company, desired to move from expensive traditional car accident damage assessment methods more cost-effective AI-based damage assessment methods. Starting from a dataset of annotated 30.000 images, we have trained various Deep Learning architectures to accurately extract damaged areas with severity estimations. The quotes generated by our system use vehicle parts segmentation coupled with semi-supervised learning plus a global car price estimate computed through a comparative analysis of several international markets. 

Semantic Analyzer for Procurement Data Cleanup. Korea’s industry-leading provider of purchasing and supply management solutions asked us to build an automated system to assist them in cleaning up the information contained in procurement databases. Such information is gathered from a large number of sources and people and is inherently ambiguous. We built a customized semantic analyzer that uses both deterministic methods and convolutional neural networks to learn how to remove ambiguity in procurement data.

Computer Vision Retina Matching. Our client manufactures laser-based eye surgery machines. We built a software module to assist the surgeon in evaluating the treatment procedure by composing a map of all the treatment points on the entire retina by applying image enhancing to the laser shooting images, identifying their position on the retina image, and placing every laser point in the right position on the retina.

Bigram-Level Keystroke Authentication. We designed an authentication mechanism through keystroke analysis based on bigram embeddings. Unlike conventional methods, our model targeted the temporal features of key bigrams (i.e., pairs of consecutive pressed keys characterized by hold latency, inter-key latency, press latency, and release latency) to generate user embeddings using a Transformer-based neural network that distinguished between different bigrams. We then employed self-supervised learning techniques to compute embeddings for both bigrams and users.

Product Recommendation System. We built a customized product recommendation system using machine learning and semantic analysis for a large European on-line retailer. The double-digit measured increase in sales attributed to the system’s recommendations proved the high value of our customized system.

Cryptocurrency Price Forecasting. Using a combination of complex time-series analysis techniques, machine learning applied to correlations of time-series parameters, and deep learning applied to the semantic analysis of news feeds, we built a price and price trend forecasting system for a Korean customer engaged in cryptocurrency arbitrage. 

Blockchain-based Software. We developed a system supporting late payments using our innovation in transactions, as well as allowing off-chain agreements between the parties involved in a transaction. The system also provides service extensions through a catalog where Blockchain members can subscribe and offer real-time services with context-based adaptive pricing. 

Deep Learning Classifier. We implemented the dual-distribution-based learning methods defined at Caltech in 2015 and improved on these methods to make them computationally efficient. Demonstrated its classification capabilities and computational efficiency on million-examples datasets for customer product recommendation.

Legal Document Comparison Semantic Analyzer. Using our own research capabilities, we extended the theory of Latent Semantic Analysis with new techniques that yields significant improvement in the accuracy of analyzing legal documents in order to classify them by similarity on specific topics and at all levels of granularity. Based on this new technology, we implemented an efficient semantic search engine to be used in comparing such documents for legal analysis purposes.

Autonomous Vision-based Drone Navigation. We designed and implemented a computer vision system for self-guided autonomous drones: by converting 2D scenes into 3D interpretations we perform semantic analysis on such scenes in order to guide the drone to complete its tasks.

Natural Language Guidance for Drones. We built a Smart Drone Assistant that controls a drone’s operation using natural language instructions: the system is capable of answering queries by analyzing 3D scenes and generating natural language constructs from aerial images. This technology allows for efficient communication between a human operator and the drone, especially when the human operator is an expert in the application domain but not necessarily an expert pilot.

Data Center Security Vulnerability Detection. We designed and built a cloud-based big data analytics platform for providing security risk assessment across an entire organization: our solution translates the organization’s security vulnerabilities into a pragmatic business perspective by analyzing all the relevant data about the organization.

Computer Network Mapping. We built a computer network topology discovery tool that allows a complete mapping of the entire computer network of an organization (hardware, software, and communications); based on a set of agents running both outside and inside the organization’s network, our tool is also capable of monitoring any changes in the hardware and software configuration and tools and is being used in both compliance and security applications.

Product Configuration System. Built the second ever product configurator for a multinational computer company using deterministic AI (expert system) technology.