Hailing from Ottawa, Canada, David Manouchehri is the Director of ai.moda SEZC and is helping to build scalable artificial intelligence (AI) and machine learning (ML) solutions from his Cayman Enterprise City (CEC) office in the heart of George Town. He’s the inventor behind some exciting new technology that is helping businesses to leverage their existing workforce, avoid the typical outsourcing risks, and radically streamline processes, while training AI/ML. We recently sat down with David to discuss new tech, the potential AI/ML training, and life in the Cayman Islands.
CEC: For those of us who don’t know you, tell us a little bit about how you got into the tech space, and what you do now?
DM: I got my start in tech as a Junior Security Analyst at the Department of National Defence (Canada). After a year or so, I moved to a small defense contractor (Linchpin Labs) to focus on Chrome and Android/Linux security. During my time at Linchpin Labs, we were acquired by L3Harris.
After the acquisition, I decided to leave L3Harris and started offering cybersecurity consulting independently for Android/Linux and Chrome based applications. Since then, I’ve branched out into the AI space as well and started ai.moda. My role is primarily as a director for the dozen or so employees and contractors we have.
CEC: What does your company ai.moda do?
DM: We have three parts of our company:
Consulting: Mostly focused on cybersecurity related services and automation. For example, we built a Terraform for our partner’s at Corellium.
Valkyrie (in development): Automated zero-trust hardening for Linux applications and cloud workloads. We mitigate and reduce the impact of vulnerabilities (including 0days) for enterprises and individuals. We plan to enter Valkyrie into DARPA’s AIxCC challenge.
YourCrowd (in development): Automate delegating tasks to both your bots and human workers with an MTurk compatible API.
CEC: And for those of us who don’t know what MTurk is?
DM: MTurk, or Amazon Mechanical Turk, is a crowdsourcing platform that connects businesses with a large, flexible workforce to perform tasks that require human intelligence. These tasks are usually simple and can range from data validation and research to more subjective tasks like surveys.
Imagine a company wants to know which of their ad designs is more appealing to the general public. Instead of conducting a costly and time-consuming focus group, they can simply upload the different ad designs onto MTurk and ask workers to choose which one they find more interesting. Within a short time, they'll have hundreds or even thousands of human responses, providing valuable insights at a fraction of the cost and time. So, MTurk essentially acts as a bridge between businesses in need of quick, human insights and individuals looking to earn money for performing these simple tasks.
CEC: How does MTurk help with developing ML and AI models?
DM: MTurk serves as an invaluable resource for data collection and data labeling, two foundational elements in the development of machine learning and AI models. Let's break it down:
Data Collection: Gathering raw data is the first step in any machine learning project. MTurk can be used to collect various types of data like text, images, or even user behavior. For example, you can ask workers to transcribe audio clips, annotate images, or fill out surveys.
Data Labeling: Once you have raw data, it often needs to be labeled for supervised learning algorithms. Labeling can be a time-consuming process, but MTurk helps speed it up. For instance, if you're developing a computer vision model to identify objects in images, you can use MTurk to have humans draw bounding boxes around specific objects or label them.
Data Validation and Cleaning: Beyond collection and labeling, the data often needs to be validated for quality and consistency. MTurk allows for redundancy in tasks, meaning multiple workers can perform the same task to ensure accuracy.
Human-in-the-Loop (HITL): Some AI models, especially in the early stages, may require human intervention for tasks that the model can't yet handle effectively. MTurk can provide a scalable way to include human judgment in these cases.
Cost-Effectiveness: Traditional methods of data collection and labeling can be costly and time-consuming. MTurk provides a cost-effective alternative by leveraging a vast, distributed workforce.
Speed: Time is often of the essence in tech development. MTurk allows for rapid turnaround times, which can be crucial when you're iterating on a model.
By providing a scalable and cost-effective solution for these critical steps, MTurk plays a significant role in accelerating the development of robust and reliable ML and AI models.
CEC: Are there any privacy concerns with using MTurk?
DM: While MTurk is a powerful tool, it's important to acknowledge the privacy concerns that come with using a public crowdsourcing platform, especially for sensitive or proprietary projects. When you're outsourcing tasks to a pool of anonymous workers, you have less control over who is accessing your data. This can pose a security risk and may not be compliant with privacy regulations like GDPR or HIPAA.
That's where our platform, YourCrowd, comes into play. YourCrowd was developed with these concerns in mind. It implements the same MTurk API, meaning you won't need to make any code changes to transition from MTurk. The key difference is that YourCrowd allows you to use your own trusted workforce instead of contracting out to untrusted third parties. This way, you maintain full control over your data while still benefiting from the scalability and efficiency of crowdsourced labor.
By offering the flexibility to use a vetted workforce, YourCrowd aims to bridge the gap between the convenience of crowdsourcing and the security of in-house data handling.
CEC: Can you explain the tech behind ai.moda’s YourCrowd?
DM: We built YourCrowd entirely on Cloudflare Workers, which provides us with a serverless computing environment that allows for excellent scalability and performance. Utilizing Cloudflare Workers has enabled us to deploy our application closer to the end-user, thereby reducing latency and improving the overall user experience (compared to AWS, which only runs MTurk in a single eastern US region).
CEC: ai.moda was recently announced as the first in the Caribbean region to be accepted into the Cloudflare Workers Launchpad funding program. Congratulations! Tell us more about the opportunity and your plans moving forward.
DM: Being part of the Cloudflare Workers Launchpad helped us a lot with improving our development speed by giving access to some of Cloudflare’s senior engineers. Additionally, Cloudflare also helped with introductions to some of their investor partner’s, which we’re currently working with to begin our Seed round of funding.
CEC: The potential of ML and AI is tremendous, and the industry is moving faster than ever. What do you do to say on top of it all?
DM: You’ve got to pick and choose what you want to be an expert in.
CEC: Before you made the decision to establish a physical business presence in the Cayman Islands, what other jurisdictions did you consider and how do they compare? Why is the Cayman Islands a great place to develop new tech?
DM: I considered three locations for ai.moda:
1. United States - Great for high skilled talent, but high taxes and visa requirements are non-trivial. Even once set up, there’s the constant headache of needing to get visas approved for immigrant workers.
2. Canada - Lower cost of living, but high taxes and a lot of constant paperwork for expenses/accounting. Similar to the US, it is non-trivial to get visas for immigrant workers.
3. Dubai - Decent cost of living, no income taxes, and easy to get visas. Unfortunately, geography isn’t on your side if you’re pulling from a western talent pool (e.g. New York, Ontario, Florida, etc.) It is not trivial to convince new hires to move across the globe.
4. Cayman - Very easy to convince highly skilled talent to relocate (it’s a <4-hour flight to NYC or Toronto), no income tax, and the visa process is easy enough that I’m able to send the paperwork to CEC without needing a lawyer to review it. Most Venture Capital (VC') firms are already used to Cayman corporations, so it’s a very familiar legal system as well.
CEC: What’s the best advice you can give someone who is considering moving to the Cayman Islands?
DM: Look at some of the existing companies and see what mutual contacts you already have.
CEC: What surprised you the most about the Cayman Islands?
DM: The ease of relocating and making connections. I made more Canadian friends here in Cayman during 2021 than I did in Canada during 2020.
CEC: When does ai.moda officially launch and how can individuals learn more and/or get involved?
DM: Individuals can go to our website and blog, which we try to keep updated, to learn more on our initiative and our launch dates!
Thanks David! We’re excited to see whats next up for ai.moda and keep us updated with how the Cloudflare Workers Launchpad goes!
Are you an innovative business on the hunt for an offshore business location? CEC is here to help. We make growing an international business from the Cayman Islands easy, affordable, and functional. Learn by visiting our business set-up page.