condense9::HarkFormerly, Teal
Build stateful and portable serverless data pipelines in record time with Hark.
  • Zero user-managed IAC.
  • Python 3.8 inter-op (import & call functions).
  • Fully serverless: Lambda and DynamoDB on AWS.
  • 60s to deploy from scratch, <2s after that. It's fast.
// Hark is a programming language & VM.
// other imports...
import(transcode, python.video_tools, 4);

fn on_upload(bucket, key) {
  if check_format(bucket, key) {
    hls = async transcode_hls(bucket, key);
    mp4 = async transcode_mp4(bucket, key);
  }
  else {
    error(“bad format”);
  }
  update_database(await mp4, await hls);
  print(“Success!”);
}

Language inter-op

Interface with databases, transform data, and do all the other hardcore business logic in Python.

Use your favourite Python tools and libraries to do the heavy lifting while Hark handles the glue, seamlessly passing data to & from Python functions.

Test Python functions as units. Test them together as Hark programs.

Portability

The Hark runtime abstracts "storage" and "compute", so any platform that can provide these, can run Hark.

The side-effect is that you can test your application locally, using local storage and compute, then deploy to the cloud. If it behaves differently, that is considered a Hark bug.

The AWS backend has already been implemented, and GCP is next.

Effortless Concurrency

Hark hides the complexity of concurrent processes in AWS behind familiar programming constructs (async/await), letting you describe systems with multiple "threads" that pass data and wait for each other.

When running in AWS, threads run in separate Lambda contexts and share data through DynamoDB, transparently handled by the Hark runtime.

Advanced Metrics

Standard output and logs across threads are automatically scoped by project and "session" -- a bit like structured logging, but built-in.

Hark Events enable function-level tracing and performance analysis.

When things break, Hark stack-traces show the entire context, not just the current thread (or Lambda function).

Get Started Instantly

$ pip install hark-lang

$ hark init

And then check out the documentation or the examples!
July 2020

Let’s talk about serverless.


Serverless has problems, but really it’s pretty great. Lots of smart people think it’s the future.

We love forgetting about servers, eliminating idle time, and scaling magically.

It's great. But we’ve traded that operational simplicity for development complexity.

Building anything non-trivial takes time, care, and stacks of infrastructure. Architecture diagrams for the most mundane data pipelines.

Web development used to be pretty simple. Everything could run locally, and everything was defined in code.

It was so simple.

What if we could have the same experience with serverless? We should be able to have simplicity in development and operations, while remaining serverless.

Condense9 is making that happen.


The best way to get invovled is to build something awesome with Hark and tell us about it. Hit the Star on GitHub. Tell us how it helped you, or how it failed to meet your expectations.


There’s so much more to come.

Get updates.

Max one message per week, just about Condense9. We respect your privacy.

Questions? Requests? Just saying hi?

→   hi@condense9.com