1/ 2020 has been one helluva year for #engineering at @GraknLabs: we rewrote #grakn and #graql. We took the best of our #design, #math and #compsci from our work over the past 5 years that the #community loved, and we #rebuilt it with the best #software #architecture and #mindset

2/ We took @RocksDB (an #opensource powerful low-level key-value storage of byte arrays, built by the engineering team at @Facebook) and we #rebuilt #grakn from scratch on top of it. @GraknLabs

3/ We built a new #hypergraph #storage engine tailor-built for #grakn's #knowledge representation system. One that's a hybrid of a persisted and in-memory #graph, and ensures thread-safety. @GraknLabs

4/ Then we built a new knowledge representation system, extending #grakn's previous expressivity with behaviours such as role inheritance, overriding attribute/role types, scoping roles to relations, all while providing better #typesafety through type restrictions. @GraknLabs

5/ We rebuilt #graql (while keeping the most of the same grammar) with a new way to represent the #datastructure #underthehood that is natively a #graph and semantically #logical. @GraknLabs

6/ We rebuilt #grakn's @grpcio-backed #rpc #server that optimises throughput and latency of #queries (especially in the #cloud) by designing the reactive flow control ground-up. @GraknLabs

7/ We built a #benchmarking system through a #simulation engine that generates a diverse dataset, and executes queries onto #grakn that hits all of our #programming #API. @GraknLabs

8/ We rebuilt #grakn KGMS (now called Grakn Cluster) to run a #cluster of Grakn database servers to provide #highavailability. We redesigned it ground-up following the RAFT protocol natively. @GraknLabs

9/ We built a new #CI & #buildsystem: Grabl ("Grakn's builder"), ensuring the reliability of our entire #software workflow from #sourcecode analysis, #tests, #tracing, #benchmarking, #cluster jobs, #release validation, #release deployment, #dependency update, and more. @GraknLabs

10/ We cannot wait to share our work with our #community very, very soon. 2021 will be an amazing year as we get to roll out all of our new technologies into the hands of #engineers in our @GraknLabs community. We'll be sharing more about how our #engineering works!

11/ Right now the team is in the midst of completing the new #reasoning engine, designed ground up using our very own #multithreaded #eventloop and #concurrent #actormodel. The new #reasoner #computation will be natively #parallel! @GraknLabs

12/ And personally, I'm in the final hours (or days) of completing the rewrite of #grakn's new #traversal engine, that has a new and more powerful #queryplanner with new #maths based on #linearprogramming, and will be natively #concurrent from the get-go. @GraknLabs

13/ We started rewriting #grakn from scratch earlier this year, not knowing for certain how long it would take to when we can run a "match query" using #graql again, and now it's about to come full-circle. And the @GraknLabs team have been working tremendously hard.

14/ Personally, for me, it's been 15hr days, 7 days a week, for 45 weeks straight. Now #grakn is about to run again, and it's #emotional. A life's work is summing up to one moment. It's like running the last KM of a #marathon you've been training for your whole life. @GraknLabs