Viktor Sanca

👨‍💻Software Engineer/Researcher in the Bay Area
👨‍🎓CS PhD from @ICepfl @EPFL
💡working on data management, systems, ML, and their complex interactions.

[Viktor Sanca @sanviktor]

@ashaikhha Congratulations! 🎉

[Viktor Sanca @sanviktor]

This is part of my vision toward analytical engines that support both structured and unstructured data through extending the relational model and existing execution infrastructure and systems: https://t.co/De3dQywCO3 (ICDE'23) and https://t.co/rqtcCcDN54 (TKDE invited issue)[5/5]

[Viktor Sanca @sanviktor]

✈️ I'm going to Santiago, Chile to attend @SIGMODConf and present our work at @DaMoN_workshop.

⚡Fast and efficient data access is the cornerstone of actionable analytical insights. But novel use-cases and vector embedding analytics challenge the current approaches.

[1/5]

[Viktor Sanca @sanviktor]

Paper will be available here: https://t.co/FINknSDQj8 Looking forward to presenting on Monday!

📩If you're working in this domain, I'd be even happier to get in touch and exchange ideas and potentially collaborate - workloads and benchmarks should propel more research!

[4/5]

[Viktor Sanca @sanviktor]

I look forward to presenting this work at @ComputerSociety @icdeconf in Utrecht, Netherlands!

#vectorDB #relationalDB #analytics

[Viktor Sanca @sanviktor]

Optimizing Context-Enhanced Relational Joins has been accepted @icdeconf'24. Collaborating with Manos and @ailamaki, we explored using vector embeddings to enable data processing engines to reason about previously opaque context-rich data.

[Viktor Sanca @sanviktor]

💡Compared with a vector index, with relational filters and different access patterns, highly optimized exhaustive processing can be beneficial! We start from a column store, composable with traditional relational operators, towards our vision of model-relational analytics.

[Manos Athanassoulis @manathan1984]

Our book on "Data Structures for Data-Intensive Applications" co-authored with Stratos Idreos (@HarvardDASlab) and @DennisShasha can be downloaded for free from the publisher for the next few days (until Feb 12): https://t.co/0lsOjSC6hy

#FNT #DataStructures #Textbook #Databases

[Viktor Sanca @sanviktor]

Vector embeddings enable data analytics beyond similarity search.

Relational operators gain new semantics with embeddings (https://t.co/De3dQywCO3) - and many new challenges to solve.

In our preprint we dive deeper into embedding-based similarity joins: https://t.co/NkTwvcSMyt

[Manos Athanassoulis @manathan1984]

#PhD admissions at #BU #CS are now open! More information for the CS PhD program can be found at: https://t.co/87ENRAEMQp

Application deadline: December 15, 2023

Apply at: https://t.co/fNDsbdkCyx

#FAQ: https://t.co/ClzZNinazr

Please RT!

[Viktor Sanca @sanviktor]

It was a joy to give a talk to the graduate class and Prof. Angela Bonifati's DB group at Lyon 1 University.
I presented my research under the umbrella of System, workload, and context-conscious analytics.
Thank you @ang3ela for the photo and the fun discussion with your group!

[Renata Borovica-Gajic @RBorovica]

I am looking for a #postdoc to work with me on #AI-driven #Databases as part of my #DECRA at #unimelb.

If you have a PhD and do research in #databases, #AI (#ML, #RL, recommender systems, user intent prediction) OR #HCI, please apply!

More details here: https://t.co/rdEA2Gb5t4

[Matthias Boehm @matthiasboehm7]

The new book "Data Structures for Data-Intensive Applications" (by @manathan1984, Stratos, and Dennis) certainly made it to my reading list: https://t.co/aac6N4KCIl

[Viktor Sanca @sanviktor]

6/ From modern hardware and its interactions - to ML and rethinking established algorithms, I look forward to several presentations at @VLDBconf #VLDB2023!

If anyone is interested, I am happy to chat or catch up at the conference or in the #Vancouver area.

[Viktor Sanca @sanviktor]

1/ I'm very excited about having 4 workshop papers accepted at @VLDBconf #VLDB2023, outlining data management research directions and challenges I am passionate about.

Research done at @dias_EPFL @ICepfl targets different parts of the data management pipeline.

[Viktor Sanca @sanviktor]

5/ Finally, we propose "Improving K-means Clustering Using Speculation" by sampling the data for fast cluster exploration, in parallel running over the full data for true clusters, merging the two phases for faster convergence at #AIDB with S. Igescu, @elenizapridou, @ailamaki.