Partnering and Collaboration
I have served as assistant professor at the University of Guadalajara and currently provide consultancy services. I have helped design and update mathematics curricula for engineering students, as well as proposed and led lines of research. I also lecture on pure and applied mathematics and computer science. My work focuses on foundational mathematics, where I developed a canonical set theory that addresses key problems such as Hilbert’s 24th Problem and Benacerraf’s Identification Problem. This theory offers novel insights in set theory, category theory, and abstract algebra, yielding practical applications in computer science. Notable innovations include a fast and scalable in-memory adder for matrix multiplication and a homomorphic encryption scheme that reduces noise, time, and energy consumption. These developments have transformative potential for current fields of research like next-generation Compute-In-Memory processor architectures, privacy-preserving AI and secure cloud computing. I am now seeking investment and collaboration to further advance and commercialize these innovations.
I have great initiative and versatility in setting up, organizing, and managing teams to execute project workloads across different systems and platforms. Apart from leading research projects, I also have experience with administrative tasks including international patent processes and basic knowledge of patent procedures, designing technical documents and websites, project management, creating and administering work spaces, etc.
What is our company going to make? The product and what it does.
We are developing a modern mathematical foundation to solve core computer science challenges, backed by global patents that enhance computing throughput, efficiency, privacy, and performance. I currently manage three key projects. The first two are aimed at solving fundamental problems in computer science and improving AI implementation: data privacy/security and computational efficiency/throughput.
I) HOMOMORPHIC ENCRYPTION (HE) BY RANDOM CHANGE OF VARIABLES: Today’s biggest challenge in cryptography is the problem of processing encrypted data—without decrypting. Methods for achieving this exist. These are called Homomorphic Encryption schemes, but they have not been fully implemented in the most important applications because of significant setbacks such as computing power requirements, noise (loss of precision), latency, and energy consumption. For this reason, HE is a topic of current R+D with industry giants such as Microsoft and IBM investing heavily in this technology. I have developed a novel “Random Change of Variables” method to overcome HE’s traditional limitations. This innovation enables secure, privacy-preserving AI, federated learning, and cloud computing without sacrificing efficiency and performance. A PoC is publicly available online, and I’ve presented the theoretical framework at international conferences. Currently I am working with my team to deploy a Minimum Viable Product that outperforms IBM’s and Microsoft’s encrypted arithmetic libraries.
II) SIMPLE AND LINEAR FAST ADDER (SLFA) FOR IN-SITU MATRIX MULTIPLICATION: I have also worked on a new design for the subunit in a processor that is responsible for executing the basic arithmetic operations. Modern processor architecture, or “Von Neumann Architecture”, separates the memory and the arithmetic unit into two distinct subunits connected by a bus that moves information back and forth between them. The Von Neumann Bottleneck refers to the bandwidth and resource demand of this bus, which is responsible for most of the processor’s energy consumption and time delay. We improve efficiency and performance by addressing this bottleneck through a Compute-In-Memory architecture. The SLFA design enables faster, low-energy matrix multiplication for AI, cryptography, and digital processing. A patent application has been filed in multiple countries, and is supported by the receiving office (USPTO) written opinion, letters of acceptance in the relevant countries, peer-reviewed research, as well as an external appraisal and Potential Buyers Identification Report. Furthermore, our Compute-In-Memory IP is a new architecture based on existing technology, unlike other CIM proposals that research for new transistor/memory types with huge R+D costs and unprofitable manufacturing processes.
III) SEMI-SUPERVISED ALGORITHMIC TRADING FIX-API: I participate in the development of a trading system based on real-time technical analysis. My roles include software architecture, testing, and debugging, in collaboration with a software developer and market analyst to optimize performance and strategy execution.
How far along are we?
I) HOMOMORPHIC ENCRYPTION (HE) BY RANDOM CHANGE OF VARIABLES: We have successfully implemented a POC that can process encrypted data for a loan application, and establishes validity of our proposed HE scheme as memory efficient, low-power and low latency. However, our PoC can only be used for evaluating an Encrypted Loan Amount Calculator. It cannot be used for any mathematical operations other than the predetermined credit score calculator. The biggest evolution between the PoC and the Minimum Viable Product, is that the MVP will allow for a wide range of functions to be evaluated in ciphertext. We aim to define an encrypted arithmetic library that competes with Microsoft’s Simple Encrypted Arithmetic Library (SEAL) and IBM’s Homomorphic Encryption Library (HELib). Our library of functions will include standard operations for encrypted numbers, vectors, and matrices including addition, dot product, fast matrix multiplication, derivatives, and other conventional operations with high value for IT businesses.
II) SIMPLE AND LINEAR FAST ADDER (SLFA) FOR IN-SITU MATRIX MULTIPLICATION: I have presented the mathematical foundation of this design at various conferences and filed for global protection under the Patent Cooperation Treaty in eight key countries (Canada, China, India, Japan, S. Korea, Singapore, UK, USA). Supporting material includes written opinion by the receiving office (USPTO), peer-reviewed research, and an external patent valuation and Potential Buyers Identification Report. The next steps are simulating, benchmarking and prototyping in collaboration with the University of Guadalajara, although we can start licensing Right of Use as of now.
III) SEMI-SUPERVISED ALGORITHMIC TRADING FIX-API: The model is being tested, troubleshooted and debugged. We are now building a transaction history in order to offer a robust model for various markets.
How long has our team been working on this?
For the past 10 years, I have focused full-time on my own independent research while remaining self-employed. The other team members each hold at least six years of professional experience in software development, modeling and/or market analysis.
When will we have a version people can use?
I) HOMOMORPHIC ENCRYPTION (HE) BY RANDOM CHANGE OF VARIABLES: The PoC was developed in four (4) months by a two-person team, and demonstrates that our method successfully addresses latency, energy consumption, and noise issues in traditional homomorphic encryption techniques. A fully functional version with market competitive features is expected to take eight (8) more months with a three-person team, and is already underway.
II) SIMPLE AND LINEAR FAST ADDER (SLFA) FOR IN-SITU MATRIX MULTIPLICATION: We estimate ten to twelve (10 to 12) months for simulating, testing, troubleshooting, benchmarking and prototyping our fast adder architecture, once the necessary funding is secured. Expenses include proprietary EDA software tools, laboratory operating costs, hardware, workstations, Senior Electrical Engineer, two Assistant Engineers, key participations at industry events, patent expenses such as translations, attorneys, application fees, maintenance fees, etc. We are preparing two patents to extend the functionality of the original SLFA adder to enable parallel matrix multiplication.
III) SEMI-SUPERVISED ALGORITHMIC TRADING FIX-API: In six months time, once we have built a track record and a solid statistical foundation for accurate projections, we will release a proprietary API for licensing. This API will help investment funds make real-time decisions based on our model.
Why did we pick these ideas to work on? What is our domain expertise in this area? How do we know people need what we're making?
I have worked on redefining mathematical foundations proving new theorems that connect various areas of mathematics and provide clearer proofs of classical results. Applying these methods to computer science has led to the development of two key innovations:
I) HOMOMORPHIC ENCRYPTION (HE) BY RANDOM CHANGE OF VARIABLES: Our novel approach to HE addresses critical challenges such as noise, latency, energy demands, and computing requirements that have hindered large-scale implementation. Unlike traditional methods, our approach doesn't rely on homomorphisms but uses alternative mathematical concepts to preserve data privacy and security during processing.
II) SIMPLE AND LINEAR FAST ADDER (SLFA) FOR IN-SITU MATRIX MULTIPLICATION: Today’s processors are facing the challenge of overcoming limitations for applications demanding massive processing power. Although processors may seem fast and efficient to us, the truth is that at the frontier of modern applications, the processing power of the most advanced technology is still insufficient, and our capability to improve them simply by making smaller transistors (Moore’s Law) is reaching theoretical limits. Our design addresses Von Neumann’s Bottleneck with a Compute-In-Memory (CIM) architecture that features a scalable circuit for efficient matrix multiplication, crucial for high-performance computing. The innovative addition algorithm has been presented in conferences and published in peer-reviewed journals.
III) SEMI-SUPERVISED ALGORITHMIC TRADING FIX-API: Our easy-to-use FIX-APIs support semi-supervised algorithmic trading with robust risk management, which is highly valued in the financial sector for maximizing profitability and optimizing positions.
Who are our competitors? What do we understand they don't?
We focus on pressing computer science problems and develop solutions based on a new theoretical foundation. My work as a mathematician revolves around finding optimal mathematical representations of numbers and mathematical structures, as well as defining operations at the most fundamental level. Through numerous internationally recognized, peer-reviewed publications and conferences, I have presented insights that provide simple solutions to previously complex problems. Key businesses like Microsoft, IBM, Google, AWS, Zama, Inpher, Nvidia, Micron, Intel, Qualcomm, AMD, and academia are working on the two main problems we are solving (Homomorphic Encryption and Compute-In-Memory). However, we see these organizations more as potential clients and partners rather than competitors.
How valuable are our products to the market?
By solving core inefficiencies in encryption and processor design, we drive innovation while creating substantial commercial value. We deliver advanced, deployable solutions to enable seamless horizontal and vertical integration in computing and IT, potentially setting new industry standards. Our enterprise-level B2B solutions and services enhance the technological ecosystem by addressing critical challenges with impact across industries.
I) HOMOMORPHIC ENCRYPTION (HE) BY RANDOM CHANGE OF VARIABLES: We have developed a software solution enabling computation on encrypted data, protected by trade secrets, patents, and technical measures. Currently, the HE market is valued at ~$200M (2024), hindered by latency, power consumption, and noise issues that do not allow for a full implementation in the most critical industries. However, our breakthrough HE scheme aims to overcome these limitations, unlocking massive market growth while positioning us as a key player.
II) SIMPLE AND LINEAR FAST ADDER (SLFA) FOR IN-SITU MATRIX MULTIPLICATION: Our hardware-level optimization for high-performance ASICs, GPUs, and TPUs using Compute-In-Memory processor architecture has been filed for protection in strategic tech hubs (Canada, China, India, Japan, S. Korea, Singapore, UK, USA) to maximize licensing opportunities with leading chip designers and manufacturers. The proposed design has the potential to set new industry standards in efficiency and performance, largely surpassing the competitor’s processor architectures, with little R+D resources invested.
III) SEMI-SUPERVISED ALGORITHMIC TRADING FIX-API: A non-patentable software solution protected through a combination of technical and legal IP safeguards and is offered for licensing.
