A Technology Convergence Story

TLDR; Looking back to the the WHY behind Heimdall. Updates on the completion of Heimdall Unstructured (Project Ymir). Announcement of the next leg of Heimdall. Follow Heimdall on X: @HeimdallML

Convergence

Heimdall was an idea that came to me while I was in grad school. This was also an exciting point in my career because from 9-5 I was still doing everything I can to stay ahead in my career. I was leading a scrum team and it was a strange experience because everyone I was leading had around 10 years of experience on me. After work, I would rush to class from 6 to 9, to learn the ins and outs of Data Science, which was really starting to interest me. After class, I was being swept away by blockchain technology. For almost two years, my life was a convergence between full stack software engineering, data science, & blockchain tech.

This was the perfect convergence which led to Heimdall. In my day job, I was guiding the technical direction of a massive project at one of the largest banks in the world. This helped me to see the technology considerations made by engineers who scaled products to millions of people. After work, I was learning everything about adopting a data driven mindset, including both theory and implementation of predictive models. After class, I was learning how there was a technology revolution occurring across the world which was mostly powered by the idea of digital ownership.

Digital ownership is the cornerstone of why I built Heimdall. Today, if you go build something in AWS or GCP or Azure, they make it so hard for you to move out of their ecosystem. Makes sense - they are there to make money. But what if there was a different way?

As Heimdall matures - this will become clearer.

Project Ymir

I am ecstatic to announce that Project Ymir has been delivered as per my initial vision. Users can now convert text, images, and audio files to tabular metrics. These metrics can then be used to build powerful models to do important things like classification or even regression. The line of products which fall under project Ymir are designated as Heimdall Unstructured. I was recently reading about how a company was taking pictures of their crops to identify when they should cut or harvest their crops. Instead of spending millions of dollars hiring a consulting team or even an army of engineers, you could take 200 images of your crops, identify which ones should be harvested and then extract the features using Heimdall Vision. You have essentially done 90% of the hard work for building a predictive model for this use case.

Project Ymir will stay in alpha stage based on usage. Heimdall is a solo project and its product direction will always be data driven. I am constantly obsessed with how users spend their time on Heimdall and what features they really enjoy. I am working on establishing some partnerships with professors in the area so I can build an internship like opportunity for students who are interested into researching unstructured data. There are some exciting updates on this effort but that's for another time.

I built out the Ymir initiative because I felt that feature engineering is a time consuming process which should be completely removed from the data scientist role. A data scientist should be able to process all their unstructured data, get tabular data, and then focus on which ones they actually want to keep for modeling. Many will disagree but this is the idea thesis for Heimdall Unstructured.

Want to see more work done on Heimdall Unstructured? Go activate API keys! Want to contribute? Shoot me an email.

Key Performance Indicators

Heimdall was born on May 17, 2023 and it will be 5 months since Heimdall's birth in exactly 2 days. During this time there has been a lot of activity and I want to take this time to highlight some of those. Heimdall is a data driven product so at the end of the day these metrics will drive all product and feature development.

• 3,253 unique visitors over a 5 month period
• Visitors from 108 countries across the globe
• 1,302 engaged users (40% of population)
• Top user interests: Technology/Technophiles & Media & Entertainment/Light TV Viewers
• 60 active users
• 37 active models
• 5 active Heimdall Unstructured API keys
• 424 followers on X
• 86.7K impressions on tweets

These numbers may not sound like much but these metrics are a testament to the work I have put in every single day for the past 5 months. I have no funding nor do I have a team. I am passionate about my mission and I look forward to growing these numbers over the next couple years!

Whats Next?

As Heimdall Unstructured is now wrapped up, it is time to look towards the next phase in Heimdall's development. Heimdall was accepted to the Charlotte Launch program. This program helps founders in the Charlotte area identify their product-market fit. The program has been awesome and I look forward to learning more about my market and how I can get them engaged.

The next two products will be focused on the product-market fit. Heimdall Forecast is the next product that is scheduled for release in about 2 weeks time. You may even see the alpha build out sooner! Forecast will give the users ability to bring in time series data and automatically get an endpoint to get forecasts. Forecast is an exciting product because it was requested and is being guided by some incredible advisors who have seen the insides of various startups. More on this another time.

The second product is a community product. This is more of an actual community building project than an actual product line. Heimdall will start launching monthly modeling challenges with a leaderboard, starting January 2024. This essentially means that every month we will release a dataset to the public community to build models with. We will also run the same dataset through Heimdall ML and then have a leaderboard to show which models perform the best. There will be some cool ways to measure this performance. We want to inspire a new generation of data scientists and provide them a way to master their skills. There will be badges and various prizes to help engage the community.

If you haven't already, join our growing community on X: @HeimdallML

Heimdall – Harness the Power of ML

Heimdall is an automatic machine learning platform that gives you (and your team) a ML backend to build incredible user experiences that are powered by machine learning. It offers quick and efficient model building with even quicker model deployment. Heimdall allows your teams to quickly ideate and build out ML products 80% faster.

Heimdall scales from weekend projects to enterprise needs. Contact us today to be ready to compete in the digital age.