Hey, I'm Win, founder of ParaQuery.

We're building a bigger query engine with up to 5x cost efficiency compared to BigQuery by using cost-efficient cloud GPUs.

Ask: Do you know someone optimizing their data warehouse? Email win@paraquery.com or sign up here!

4x'ing BigQuery efficiency with ParaQuery: https://www.youtube.com/watch?v=uu379YnccGU 

The Problem

Data is the new oil and you're feeding your products with a bunch of it. But to do so, you need to build, maintain, and optimize your data infra. That's either a huge timesink, or you're bleeding cash to ignore it, or someone is breathing down your neck about this month's cloud bill. Why can't we just ship features?

Whether it's those 20-minute+ Presto queries, or your exploding BigQuery bill (which happens to be tracked in real time), your data infra is letting you down.

Our Solution

ParaQuery runs on highly-parallel, throughput-optimized cloud-available hardware, perfect for data processing workloads: GPUs!

Our system abstracts out the infra required to serverlessly accelerate your logic. With the extreme efficiency of GPUs, it's half the cost and double the performance of competing solutions like Databricks and BigQuery, without any platform commitments.

For SQL-based data warehouses like BigQuery, you can use ParaQuery with minimal changes to your queries. And since it's fully Spark-compatible, there's no vendor lock-in. It integrates with literally anything which has a Spark connector, on any major cloud. We can even deploy on GPU clouds.

ParaQuery makes data processing as simple as specifying:

1. Inputs
2. Output
3. Logic
4. Run!

Why?

I've been looking into GPUs and parallel programming for over 12 years – starting out with crypto. I left that field pretty fast, but I never stopped thinking about efficient compute and GPU acceleration. Even more so when, years later, I had to wait for my laptop to finish 18-hour physics simulations. And during my professional career, I've also had to wait on my fair share of 20-minute+ analytical queries.

In fact, I hate waiting. ParaQuery is productizing an arbitrage of hardware efficiency, and I'm not waiting any longer to do so.

Ask

1. If you know anyone who is running any big data processing scenarios (including MLOps), I'd love to learn more about their use case and see if we're a fit!

2. If you want to try it out, sign up for our waitlist at www.paraquery.com, or just email me (win@paraquery.com)