Welcome to Data Brew by Databricks with Denny and Brooke! In this series, we explore various topics in the data and AI community and interview subject matter ex
๐ง Watch the full Data Brew episode:
๐น Apple โ https://t.co/bzkIRQ2Tzk
๐น Spotify โ https://t.co/m0TjR8Hf4K
๐น YouTube โ https://t.co/PefguQxyzI
#DataBrew by @Databricks is hosted by Brooke Wenig and @DennyLee. โ
#databricks#rewardhacking
In our latest episode, @TravisAddair explains how curriculum learningโstarting with simpler examplesโand scoring methods like partial credit can help steer AI models toward correct solutions, where reward hacking is a risk.
Full #DataBrew episode โ links in the reply โคต๏ธ
The โone model to rule them allโ era is fading.
On the latest episode of the @databrew_db podcast, our CTO & Co-founder @TravisAddair, and show hosts @dennylee and Brooke Wenig, sat down for some hot coffee and tea and to chat about why #specialized AI models are winning and how reinforcement fine-tuning (#RFT) is making open-source models both smarter and easier to productionize.
In this episode, youโll hear:
๐ค Why teams are shifting from a single frontier model to a portfolio of experts
โ A clear breakdown of RFT vs. SFT and where verifiable #rewards shine
โพ๏ธ How #GRPO simplifies training vs. traditional PPO
๐ฅ Practical tips to avoid reward hacking
๐ง Real-world RFT wins in #codegen, reasoning, and #agentic AI
๐ Serving at scale using multi-LoRA to route 100s of specialized models on a single GPU
If youโre an AI builder, you won't want to miss this episode:
๐น Apple โ https://t.co/BH9lxvHq64
๐น Spotify โ https://t.co/hADK0JbrU1
๐น YouTube โ https://t.co/oZHe1ZhFMr
๐ฃ New Data Brew Episode!
We talk with @TravisAddair (@predibase) about how their new reinforcement fine-tuning platform is changing the way we create specialized AI.
๐น Apple: https://t.co/bzkIRQ2Tzk
๐น Spotify: https://t.co/x6P55LqTqV
๐น YouTube: https://t.co/PefguQxyzI
When it comes to enterprise AI, thereโs a gap between research and real-world deployment.
Databricks Staff Research Scientist Dipendra Kumar and Staff Software Engineer Alnur Ali discuss the challenges of real-world AI (messy data, security, stability), new tools being developed, and the collaboration between researchers + developers.
โ Catch the full episode of @databrew_db here: https://t.co/ok34rnB2No
@dennylee
Donโt miss this one at #DataAISummit (June 9-12)! ๐
If youโre interested in agentic AI and practical tools for developers, be sure to catch this session!๐ Check out the details below.
๐ฅ Sneak peek: Data Brew episode with Kilian: https://t.co/g8pDbjQWQB
#databrew#databricks
๐ #DennysPick ๐
I recently had the chance to virtually sit down with @KLieret from Princeton University on @databrew_db (check it out: https://t.co/15REF1STlI), and wow - heโs doing some seriously cool work! We geeked out over SWE-bench and SWE-agent, two groundbreaking tools that are pushing the boundaries of what AI can do in software engineering. Now, Kilian is taking the stage at Data+AI Summit to go even deeper, and I canโt recommend his session enough.
Why am I excited about this @Data_AI_Summit session?
๐งฉ Get an inside look at SWE-bench, the new benchmark for evaluating AI on real-world, large codebase tasks
๐ Hear Kilianโs honest take on what todayโs models can (and canโt) doโno sugarcoating
๐ ๏ธ See a hands-on demo of SWE-agent, and learn how to build and test your own software engineering agents
๐ Walk away with practical insights into the challenges ahead and how you can start experimenting with agentic AI in your own workflows
If you want to see where agentic AI is headed and how you can be part of it, add this session to your agenda!
Check out the full list of my picks โก๏ธ https://t.co/MPShPORZm5
๐Join us in San Francisco, June 9-12!
๐ Register today: https://t.co/N58pWVMU95
โ๏ธ Want to join DAIS at half price? Comment "#DennysPickโ and Iโll DM you the discount code!
#ai #data #agenticai #sweagent #softwareengineering @databricks
.@DipendraMisra and Alnur Ali explore the challenges of applying AI in enterprise environments and the tools that are helping to bridge the gap between cutting-edge research and real-world deployment.
โOne of the biggest challenges in making LLM-powered systems real is scale, both in terms of context size and high request volume.โ
๐๏ธ A new episode of @databrew_db is out! Hosted by Brooke Wenig &
@dennylee. Watch now: https://t.co/DJLa2Sh00b
๐ง Watch the full Data Brew episode:
๐น Apple โ https://t.co/8hLzsC0642
๐น Spotify โ https://t.co/JHytVkQaxS
๐น YouTube โ https://t.co/zqxtGtE6U5
Data Brew by @Databricks is hosted by Brooke Wenig and @DennyLee.
#databrew#databricks#benchmarking#DIBS
โBenchmarks should reflect the kinds of things customers actually care about.โ
In this clip, Pallavi Koppol shares why @databricks, created DIBS, a domain-specific benchmark designed to evaluate models on real-world tasks like RAG, text-to-JSON, and function calling.
๐ง Listen on your favorite platform:
๐นApple โ https://t.co/8hLzsC0642
๐นSpotify โ https://t.co/i721KVzMmi
๐นYouTube โ https://t.co/zqxtGtE6U5
Data Brew by @databricks is hosted by Brooke Wenig and @Dennylee.
#databricks#llm#enterpriseai
๐ New #DataBrew episode: Benchmarking LLMs for Enterprise AI
Pallavi Koppol (@Databricks) breaks down why domain-specific benchmarks matterโand how the Databricks Intelligence Benchmarking Suite (DIBS) helps evaluate models on real-world tasks like RAG & function calling.
๐ฌ โThereโs a shift from โone model to rule them allโ to smaller, task-specific models. General LLMs perform wellโbut for enterprise, fine-tuned models offer better accuracy at lower cost.โ
โ Dipendra Kumar, Staff Research Scientist at @Databricks#databrew#databricks
In this episode, Dipendra Kumar and Alnur Ali (@Databricks) explain why high-accuracy, fine-tuned models are essential for enterpriseโand how tools like QuickFix help bridge the gap between research and real-world deployment.
Check out the full episode! โฌ๏ธ
๐ง Apple: https://t.co/KqvSoBVbWI
๐ง Spotify: https://t.co/DjWtahafpE
๐ฅ Youtube: https://t.co/g8pDbjQWQB
Data Brew by @Databricks is hosted by Brooke Wenig and @dennylee. โ
๐ฃ ๐ก๐๐ช #๐๐ฎ๐๐ฎ๐๐ฟ๐ฒ๐ ๐๐ฝ๐ถ๐๐ผ๐ฑ๏ฟฝ๏ฟฝ!
In this episode, Kilian Lieret (Research Software Engineer) & Carlos Jimenez (Computer Science PhD Candidate) at @Princeton dive into SWE-bench & SWE-agent, two cutting-edge tools for evaluating & enhancing AI in software engineering.
๐ Curious about how QuickFix works under the hood?
Check out this new blog to explore how fine-tuning with your own data is driving better coding assistance and enterprise-ready AI. https://t.co/Jmta5iEXqh
#databricks#databrew#finetuning#llms
๐ก๐ฒ๐ ๐๐ฎ๐๐ฎ ๐๐ฟ๐ฒ๐ ๐๐ฝ๐ถ๐๐ผ๐ฑ๐ฒ ๐๐๐๐ ๐๐ฟ๐ผ๐ฝ๐ฝ๐ฒ๐ฑโ๏ธ
In this episode, Dipendra Kumar & Alnur Ali (@Databricks), discuss the challenges of applying AI in enterprise environments & the tools being developed to bridge the gap between research & real-world deployment.
LISTEN NOW โคต๏ธ
๐ง Apple: https://t.co/l7QOIICFaM
๐ง Spotify: https://t.co/dWzQ5nZJpW
๐ฅ Youtube: https://t.co/nVb6YSlA27
Data Brew by @databricks is hosted by Brooke Wenig and @dennylee.
#databricks#databrew