About the role

Recall.ai’s mission is to make it easy for developers to work with conversation data.

Conversations have become the world’s largest dataset. Every year there are more words spoken over phone calls & video conferences than on every webpage on the internet.

Recall.ai is the foundational infrastructure enabling 400+ companies to tap into this dataset.

Build Foundational Infrastructure for the World’s Largest Dataset

You'll be our seventh engineering hire in our downtown SF office. This role is in person five days a week. You'll have full ownership over major features, play a huge role in creating the culture of our engineering team, and will work closely with the founders.

You will develop high performance real-time code, deploy to large scale infrastructure, and process massive volumes of data. To give you a sense of the problem space you’ll be working in:

We process over 1TB per second of raw video at peak load
We launched 6 million+ EC2 instances in the last month

Responsibilities

Build the core product, from APIs, to real-time video processing pipelines. Be independent and own major features through development and deployment.
Operate, scale and extend our extensive video processing infrastructure. You'll be responsible for running your code in production.
Work directly with customers to solve their problems. All team members participate in customer support in order to drive empathy for our users and help us build better products.

Your Qualifications

4+ years of software development experience
Deep customer empathy, including the desire to speak with customers and make product decisions
Strong ability to learn new technologies and be productive in unfamiliar domains
Bonus points: Previously founded a startup. Extra bonus for a dev-tools startup.

You might be the right fit if…

You love writing code, but you love having impact more: We’re a team of engineers at heart, but our #1 goal is building the best possible product. That means making pragmatic choices and looking for 80/20 solutions.
You’re great at “figuring it out”: Recall.ai is a low-structure, high-trust environment. There’s minimal process, and we’re growing so fast that circumstances are constantly changing. If you can take a high level goal and independently figure out how to achieve it, you’ll fit in great.
You don’t give up: The most impactful problems we’ve solved initially seemed impossible. We work in a uniquely hard problem domain, and every important problem needs heaps of persistence.
You want to be a founder: Joining early at a fast-growing, high-talent-density startup is the best thing you can do if you want to be a founder. Anything you want to learn about startups or the business, we’ll teach you.

You might be the wrong fit if…

You’re looking for a chill gig: Startups are very hard, especially if they’re growing fast. You’ll have a ton of responsibility, and there’s always an enormous amount of stuff to do. Raw hours actually make a big difference in the impact we can make right now.
You want to ship perfect code: Velocity is a startup’s only advantage. We don’t have the luxury to wait for perfection, and we’re always looking for ways to get more done with less. We release as soon as we can, and get customer feedback immediately.

About the company

Recall.ai is building a new hyperscaler for the world's largest dataset: conversation data. Our mission is to get all the conversation data in the world into a form that's accessible and understandable by AI.

Companies like Salesforce, HubSpot, Datadog, Rippling, Deel, Monday.com, Calendly, Asana, and Workday already work with us. Every software company of note is either already powering their AI features through Recall, or in a deal cycle with us.

They choose us because this problem is genuinely important and hard to solve.

Here's why our mission is important: 50-70% of an office worker's day is uncreative toil that AI can alleviate (e.g. sending emails, scheduling, drafting contracts, etc.). But AI is bottlenecked by context, not intelligence. Just like how humans get most context from conversations, we’re giving AI the eyes and ears to access the same conversation data set.

This is a brutal infrastructure problem because conversations happen all at once. At 9am Pacific, millions of people across North America click "start meeting" within 60 seconds of each other, every single hour. Our load graph spikes vertically at the top of every hour. Our weekly peak load is 4,000% higher than our weekly minimum. When Amazon built AWS, they built it to handle Black Friday, one day a year where millions of people hit their systems at once. We deal with the equivalent of roughly 240 Black Fridays a week.

We run nearly 100,000 computers in our cluster, processing the equivalent of 3,000 full length movies of raw video per second at peak load, and we're just getting started. All the words ever written in human history are still less than 20% of one year of US office conversations.

What makes this exciting is how early we are. We captured 0.02% of office conversations last year, just in the US. If in 5 years, 20% of tech forward companies want conversation context inside their AI tools, that's a thousand times more data than we process today.

Backend Engineer [In Person, SF]