Search Engine for Qatalog
- Date: 2020Q2
Qatalog is building the WorkGraph. It aggregates data from all your companie's data silos into a single source of truth. When different teams use different tools, it's very hard to collaborate. For project management, one team may use Asana, another JIRA, and another Trello. It's hard enough keeping these things in sync, but what if you're an outsider? How can you even learn that the information you're looking for lives in this specific SaaS provider? That's the problem Qatalog is solving. Discovery and aggregation of work knowledge.
A UI with full-text search, supporting result clustering, filtering and sorting. Search results include data from integrations (e.g. JIRA).
I've also built a synchronisation algorithm from Postgres to ElasticSearch. When there's a change in one of the Postgres tables, a sync job is triggered. The job denormalizes the entity, and upserts it into ElasticSearch. This setup yielded split-second results and made for really good user experience. The final code was modular and extensible, making it easy to adapt it to schema changes and addition of new tables.
Point of Contact: Tariq Rauf (CEO)
Fund Managing Platform for Crypto Assets
- Date: 2020Q1
Who do you trust with your money? Traditionally, in order to have an advisor, you have to trust him or her with your money. But when it comes to cryptocurrency, you don't have to do that. The entire idea of cryptocurrency is to make everything transparent. But what if you could still have someone help you manage your money, but still keep it in your possession?
This platform is built to offer Fund Managers and Crypto Asset Holders an elegant solution. The solution consists of a web app that handles all the necessary functions to manage crypto funds in an easy, transparent, and secure way.
Point of Contact: Gunar Gessner (Founder)
Machine-Learning Email Classifier for Invisible
- Date: 2019Q4
- Technologies: TypeScript / GraphQL / React / Node.js / Google AutoML
Invisible Technologies allows you to delegate work so you can focus on the real work. I have used their services myself and I can vouch for them. One of the coolest things is being able to delegate over email. You see, during onboarding, we get to pick an avatar. Mine is called Donna. So whenever I have work to delegate, I email email@example.com. This email needs to be then routed to the respective team. Is it a data task? Is it a lifestyle task? Or is it just spam?
So I have built for them an email classifier using Google AutoML. Clients would send requests to the email inbox. A worker would pick an email from the queue, and run it through a previously-trained ML model. The model outputs labels, which get displayed as suggestions to Operators, who would use those labels to help them route emails to the right team. I wasn't able to completely automate the process, but I've increased operator efficiency by 10x.
Point of Contact: Francis Pedraza (CEO)