Tablingos.

Making data clean and accessible

Data has revolutionized the world. With the exponential growth of digital technology, we’re now able to obtain vast amounts of information in ways that were previously unimaginable. From the largest enterprises, down to the smallest startups, everyone deals with data.

This data is in constant motion between a distributed network of systems. For example, your business may own a sales portal where customers are able to purchase your products. Your sales would then need to be shared with your billing platform in order to charge these customers. A third service could then take this information and perform some financial reporting.

Data also needs to be migrated when changing systems. Let’s say you’re a grocery store using inventory management system A, but you heard inventory management system B is cheaper and has more features, so you’re looking to make the switch. In order to do this, however, you need to take all of your product data currently stored on system A and migrate it over to system B.

These are core processes that are constantly taking place in our data-driven world.

Organizational data is almost always riddled with errors and inconsistencies between systems that reduce this valuable resource into nothing but pain and suffering. Every system has its own expectations of how its data has to be formatted and defined, and when these requirements aren’t met, disaster strikes. Projects are delayed and money is wasted.

Consider the common challenge of customer onboarding, which often involves dealing with headaches such as data format and quality issues. When these problems arise during the onboarding process, they can result in significant delays that stretch project timelines by several months, leading to wasted time and resources. Such challenges create unwarranted obstacles in providing a smooth customer experience.

Global healthcare organizations are no better, often consisting of various disjointed systems with different ways of storing patient information, ultimately leading to a failure to effectively share and streamline this data, literally endangering lives.

This is frustrating, but it doesn’t have to be. What we need is a seamless way to validate, transform, and integrate today’s data between the systems that power our world in an efficient and automated manner.

At Tablingos, I've built something that does that. My mission is to pioneer innovative solutions that guarantee clean and useable data, enabling people and businesses to prosper in a data-driven future.

Tablingos enables businesses to define validation and transformation rules for their data. Additionally, users will be able to upload their tabular data through the app (in the form of CSVs, TSVs, etc.), which will automatically transform and validate the data according to the defined rules.

Additional methods of inputting data, such as APIs, Webhook events, FTP, and S3 events, will be introduced to enhance the initial version beyond direct file uploading. Further improvements will also be made by adding support for additional forms of data, such as JSON, images, EEG, and more.

CarNet.

The Future of Connected Autonomous Cars

Driving remains one of the most perilous activities, with 42,514 fatalities reported in the US due to motor vehicle crashes in 2022. This high risk stems from the fact that drivers, whether human or automated, typically operate in isolation. They rely solely on their own senses and reactions, leading to delayed responses and increased accident rates.

Consider the challenge of navigating around a slow-moving truck. Often, drivers must make split-second decisions without knowing the exact intentions of other road users, increasing the likelihood of collisions. Imagine if vehicles could share their plans and intentions with each other in real time. For instance, if a car planning to switch lanes could inform nearby vehicles, accidents could be prevented by allowing others to respond accordingly.

This is the core idea behind CarNet, a project designed to enhance autonomous driving through improved vehicle communication. CarNet aims to demonstrate how vehicles can publish their lane-change intentions and respond to messages from nearby devices.

The CarNet system utilizes a Sphero RVR robot equipped with a smartphone to capture live video and follow a designated lane. The robot’s movement is guided by a custom algorithm that calculates deviation from the lane and adjusts its trajectory to stay on course.

When encountering an obstacle, such as a pair of scissors, the robot broadcasts this event via a bidirectional socket to a laptop. The laptop then mimics the response of an oncoming vehicle, either allowing the robot to navigate around the obstacle or denying the maneuver if a collision is predicted. object detection is powered by a pre-trained YOLO model, which accurately classifies common objects to ensure reliable obstacle recognition.

While the prototype uses web sockets for real-time communication, the long-term vision for CarNet includes implementing Dedicated Short-Range Communication (DSRC) combined with GPS. This approach would allow vehicles to share critical information only with nearby units, improving both safety and traffic efficiency.

CarNet represents a step toward a future where connected autonomous vehicles enhance road safety and streamline traffic flow, addressing the fundamental issue of disconnected driving.

Pocket Prosthetic.

Democratizing Prosthetics

While working at Holland Bloorview, I learned about the prosthetic creation process. One of the most painstaking parts of this process was creating the mould or 3D representation of the residual limb. There were two key methods they went about doing this:

1. Fitting plaster around the patient's residual limbs: This is a painstakingly long process and can be uncomfortable for the patient as it requires them to sit still for extended periods, which can be difficult for young patients and those with mental disabilities. Furthermore, the effectiveness of this method is dependent on the skill and precision of the moulder.


2. 3D scanning: While an improvement over the previous method, the IR/LiDAR scanners used for this process are expensive and hard to come by at a local hospital. Furthermore, many sensors flash lights at the patient which can be life-threatening for epileptic patients.

That's where Pocket Prosthetic comes in. My solution makes use of the LiDAR sensor available on most iPhones today. This app allows patients to submit LiDAR scans, images, and other necessary information defined by the clinic for their patient about their residual limb and prosthetic requirements directly from their phones. This data is sent to the clinic for prosthetic generation. This approach significantly reduces the need for frequent clinic visits, making the process more accessible and convenient for patients, especially those who live far from medical facilities.

The app was developed using Swift for iOS, leveraging the LiDAR capabilities of modern iPhones to capture precise 3D models of the residual limb. The TypeScript backend is designed to handle the upload and processing of these scans, ensuring they are available to prosthetists for review and prosthetic design. The backend is hosted on AWS, and all data is stored on S3 and MySQL.

The app was one of the winning projects at Hack the 6ix, Toronto's biggest hackathon.