00:02Hello, I'm John. For the last six months, I've been working as a four-step developer at JotCapital, working on
00:10Ninja, the platform at the heart of the company.
00:14So let me show you what it is and the thing I built.
00:20JotCapital is a European private equity company that backs growth-stage deep tech companies. The hard part of the job
00:29is finding the right company early in a market with millions of players.
00:36So what Ninja does, it is a Jot's own AI platform built since 2016, watching over 5 million companies and
00:488 million patents in real-time. Today, over 75% of the deals in JotCapital's portfolio were sourced by Ninja.
01:00So Ninja has two hubs, and I work across both of them.
01:07So the front-end monomy is a React app full of interactive maps, graphs, and dashboards.
01:15The back-end Cusa runs on serverless on AWS with a Neo4j graph database that turns millions of scattered data
01:28into a connected map of the tech world.
01:35So my first project was to modernize one of the most important pages of the front-end, our advanced search,
01:44the team's main tool for finding companies, with over 20 filters across the metrics we track in our database.
01:52The old version was outdated and clunky, and it was getting in the team's way.
01:59So working directly with our product manager and designer, I rebuilt the whole page in the React front.
02:06Now it has filtered auto-completion, powered by static, matching, and regular expression.
02:14I let the team save group queries and share their queries with their teammates.
02:20Under the HOT, I rewrite every query so the new search is noticeably faster than the old ones.
02:27So it's now in production, fully replacing the legacy page, used by the investment team every day.
02:36The second project was full-stacked and touched every user.
02:41We replaced our old password login with a passwordless code-based sign.
02:52So instead of a password, you enter your email.
02:56We sent you a one-time code, you typed it in, and you are in.
03:01I built it across the whole stack.
03:04The sign-in screen on both web and mobile.
03:08The backend on AWS Cognito, where Lambda functions generate the code, send it by email, and verify it.
03:16With the code stored in DynamoDB.
03:19The interesting part was making it genuinely secure and robust.
03:24Code that expire, a limited number of attempts, and rate limiting so no one can be spamming or brute force.
03:34And now it's live as a default way for everybody to log in to Ninja.
03:42So one word about AI in a real working environment.
03:46For me, the AI coding agent, for example, cloud code, went from being a reference tool to doing a real
03:55share of the work.
03:56Concretely, it's what let me move fast in a large production code base that I didn't know.
04:03And shipped both of these projects faster than I could have done alone.
04:12So in six months, I rebuilt Ninja's search engine, replaced its login with a secure passwordless system,
04:20and I shaped how I worked with AI and with the team across the whole stack.
04:27I hope that gives you a clear picture of what those six months look like.
04:33If you have any questions, I will be happy to answer them.
04:38Een illustration of information and updates.
04:39If you don't have no agreement, we will hold in front of you and do races within the coalition.
04:41Thank you very much.
Comments