Rohan Padhye

My research group has a new website! cmu-pasta.github.io

On this week's episode of the BugBash podcast, @rocallahan tells the story of how the rr debugger came into being, enabling time-travel debugging on Firefox. No giveaway this week, rr was enough of a gift to devs everywhere. If you've ever used GDB, rr, PyTrace, or one of their many cousins, you'll want to give this one a listen!

My students wrote a blog post (pastalab.org/spaghetti-benc…) on this problem showing where AI agents struggle with concurrency, using examples such as WorkStealQueue and Kafka's DefaultStateUpdater. Feedback is welcome!

Link: pastalab.org/spaghetti-benc…

Ever used AI to fix tricky race conditions and flaky tests? Not pretty, is it? Check out "Spaghetti Bench 🍝: A SWE-Agent Benchmark for Concurrency Bug Tasks" Turns out it's a HARD problem on its own, but can be made easier with access to deterministic replay tools like Fray!

ACCEPT with Onerous Revisions

We're excited to announce our second keynote speaker for the #FUZZING'26 workshop (part of @NDSSSymposium @ San Diego, CA, USA): Sergej Dechand, CEO and Co-Founder of Code Intelligence (@CI_Fuzz)! He will present the fuzzing road ahead of us: "Where the Fuzz Are We Going?"

@michaelhilton It's so sad, spreading the propaganda of Big Jacket. Clearly 0% hot is at 0°C because if it is any hot it should be possible to have a beer outside. 100°F is just a warm afternoon, nothing special there.

@moarbugs 🤣🤣

We're excited to announce our first keynote speaker for the #FUZZING'26 workshop (part of @NDSSSymposium @ San Diego, CA, USA): Mathias Payer (@gannimo), security researcher and professor at @EPFL_en! He will present about: Advancing from "What the fuzz?" to "All the Fuzz!".

I'm excited to announce the public release of KISS Agent Framework — an open-source AI agent framework built on one principle: Keep It Simple, Stupid. Since the API is stupidly simple, I can quickly vibe code up agents and evolutionary algorithms just by providing the README dot md and a new idea. More in the blog post at lnkd.in/gaX3dZ7Q. After a month of development, I'm making this framework available to the community. Here's what makes KISS different: 🎯 Simple Architecture A clean ReAct loop implementation that you can understand in minutes, not hours. No hidden complexity, no black boxes. 🔧 Native Function Calling Seamless tool integration with OpenAI, Anthropic, Gemini, Together AI, and OpenRouter (400+ models). Your tools just work. 🧬 GEPA: Prompt Evolution Genetic-Pareto optimization that evolves your prompts through natural language reflection. Based on recent research showing this can outperform RL. 🔬 KISSEvolve: Algorithm Discovery LLM-guided mutation and crossover for evolving code. We've used this to discover faster sorting algorithms from bubble sort. 📊 Built-in Observability Automatic token & budget tracking Trajectory saving & visualization Docker isolation for safe execution The framework supports: SWE-bench Verified benchmarks onboarded AlgoTune Verified benchmarks onboarded RAG with in-memory vector search Multiprocessing for parallel execution Why open source this? Because the AI agent ecosystem needs more tools that prioritize clarity over cleverness. If you can't understand how your agent works, you can't debug it, improve it, or trust it. Check it out: github.com/ksenxx/kiss_ai I'd love feedback from the community. What features would you find most useful? #AI #MachineLearning #OpenSource #LLM #AgentFramework #Python #ArtificialIntelligence #SoftwareEngineering

Next year, SPLASH (@splashcon) and ISSTA (@issta_conf) are going to be co-located in Oakland, CA! 🥳 If you would like to run a workshop in PL/SE/Testing, submit a proposal by Jan 10th: conf.researchr.org/track/splash-i… Please RT for visibility. Contact @zengola or me for questions.

Want to shape the SPLASH/ISSTA 2026 conference program with an exciting workshop? Now's your chance: check out the call for submissions, deadline January 10th, 2026 (AoE): conf.researchr.org/track/splash-i… #SPLASH2026 #ISSTA2026

@JAldrichPL Following because we discussed this in the Program Analysis class and I'd like to hear some recent takes.

Verification folks: when building a tool like Dafny or Verus, is there a strong technical reason to prefer backwards reasoning/weakest preconditions vs. forward reasoning/symbolic execution? Or are they more or less equivalent and it's just a matter of taste?

Youtube: youtu.be/oF7krd0TQks Spotify: open.spotify.com/episode/2PH7IP… Apple Podcasts: podcasts.apple.com/us/podcast/fro…

Check out the latest episode of the #BugBash podcast! I had a great time chatting with David Wynn about automated testing in academia vs industry. Thanks for having me on!

@JAldrichPL I'm usually able to ignore small caption text at the bottom an otherwise large presentation / talk screen. But social media shorts with huge text captions front and center are insanely annoying, and I don't understand why that trend exists.

I attended a talk today & the live captions were incredibly distracting. I had a very difficult time focusing on the talk because the words were constantly appearing and my eye was drawn to them. I wonder how disability researchers think about this?

Some people say live captioning in talks are like "curb cuts"--an accessibility improvement that helps everyone. Except it doesn't! It actively harms some people (including me).

@shrey_twr Congratulations! Well deserved.

Honored to be awarded the Hima and Jive Fellowship by CMU’s Software and Societal Systems Department. Grateful to my advisors, mentors, and peers for their support along the way! s3d.cmu.edu/news/2025/grad…

It's a shame that due to the government shut down, nobody is working on advancing the bipartisan "Sunshine Protection Act" to restore mental health, reduce traffic accidents, and eliminate nasty software bugs.

The recent AWS outage incident report shows how a simple race condition involving an atomicity violation (time-of-check to time-of-use) can bring down large swathes of the internet. Test your concurrent code systematically with tools like Fray! github.com/cmu-pasta/fray

Finally some new material for motivating slides. The Crowdstrike ones were getting a bit old.

@awsTO And then shatter their worldview by introducing liveness analysis.

@awsTO Or symbolic execution. You won't know which one until they jump backwards!