Director, Core AI, IBM. Chief Architect https://xitter.cf/t.co/lo3JOcuDFA . Founder, Red Hat AI Innovation Team. PI @MITIBMLab. ❤️ Density Ratios.
Cambridge, MA
Joined July 2009
- Tweets648
- Following1.2K
- Followers1.3K
- Likes2.9K
Presenting at ICLAD 2026 (Stanford, Jul 30–31): general-purpose coding agents, no hardware-specific training, optimizing HLS designs. Decompose → ILP-assemble → scale N expert agents on the best candidates. 1→10 agents = mean 8.27× speedup across 12 kernels, >20× on streamcluster.
Interested in creating your own agent factories for hard problems, check this out: github.com/akashgit/remote-f…
w/ @Abhi_1092, Mihir Choudhury, Ruchir Puri
arxiv.org/abs/2603.25719
MIT News wrote up this work — how GIFT bootstraps image-to-CAD training data from the model's own failures (execute the code, score with geometric feedback, repair near-misses into supervision).
news.mit.edu/2026/turning-2d…
Two papers from our team at #ICML2026 🎉
1️⃣ ePF: fixes premature exploitation in particle filtering for inference-time scaling — up to 50% relative gains on hard math.
📄 icml.cc/virtual/2026/poster/…
2️⃣ GIFT: image→CAD program synthesis using a CAD kernel as verifier to bootstrap training data — matches heavy test-time scaling with 80% less compute.
📄 icml.cc/virtual/2026/poster/…
Congrats to Giorgio Giannone, friends and collaborators from @xukai92 and @_faezahmed 's amazing teams!
At ICML? Working on inference-time scaling, test-time training, SWE harness design, or engineering/math/science discovery? DM or come find me at the posters — let's chat 🙌
Two papers from our team at #ICML2026 🎉
1️⃣ ePF: fixes premature exploitation in particle filtering for inference-time scaling — up to 50% relative gains on hard math.
📄 icml.cc/virtual/2026/poster/…
2️⃣ GIFT: image→CAD program synthesis using a CAD kernel as verifier to bootstrap training data — matches heavy test-time scaling with 80% less compute.
📄 icml.cc/virtual/2026/poster/…
Congrats to Giorgio Giannone, friends and collaborators from @xukai92 and @_faezahmed 's amazing teams!
At ICML? Working on inference-time scaling, test-time training, SWE harness design, or engineering/math/science discovery? DM or come find me at the posters — let's chat 🙌
Akash Srivastava retweeted
come check out poster #5518 at NeurIPS morning session today to learn about how you can encourage diversity / prevent early-pruning during inference-time scaling and boost the performance of any model without additional training!
[1/x] can we scale small, open LMs to o1 level? Using classical probabilistic inference methods, YES! Joint @MIT_CSAIL / @RedHat AI Innovation Team work introduces a particle filtering approach to scaling inference w/o any training! check out probabilistic-inference-scal…
What does it take to scale AI beyond the lab?
At #RedHatSummit, @ishapuri101 and I spoke with Red Hat CEO Matt Hicks & CTO Chris Wright on inference-time scaling, open infra (LLMD), and making AI affordable for enterprise.
🎧 youtu.be/mj1dwrPfvb4
#NoMathAI @RedHat_AI
🚀 How is generative AI transforming the way we design cars, planes, and entire systems?
In Ep 2 of No Math AI, @ishapuri101 and I chat with Dr. @_faezahmed (@MIT DeCoDE Lab) about how AI boosts creativity, cuts design time, and works with engineers—not against them.
How is generative AI reshaping engineering design?
In Episode 2 of No Math AI, hosts Dr. Akash Srivastava (@variational_i) and MIT PhD student Isha Puri (@ishapuri101) sit down with Dr. Faez Ahmed (@_faezahmed) from MIT DeCoDE Lab to explore just that. 👇
SQuat: KV-Cache for making reasoning models go 🚀
📄paper: lnkd.in/emKhAVZu
💻 code: lnkd.in/e8TJ7N3R
From my awesome collaborators @RedHat_AI
Akash Srivastava retweeted
Excited to share our preliminary work on customizing reasoning models using Red Hat AI Innovation’s Synthetic Data Generation (SDG) package!
📄 Turn your documents into training data for LLMs.
🧵👇
Akash Srivastava retweeted
had a great time giving a talk about probabilistic inference scaling and the power of small models at the IBM Research ML Seminar Series - the best talks end with tons of questions, and it was great to see everyone so engaged : ) youtube.com/watch?v=--3rsQwM…
Come along and help us build reasoning in small LLMs
🚀 Exploring LLM reasoning—live!
We, the @RedHat AI Innovation Team, are working on reproducing R1-like reasoning in small LLMs without distilling R1 or its derivatives.
We’re documenting our journey in real-time:
🔗 Follow along: red-hat-ai-innovation-team.g…
Excited to share our latest work with @ishapuri101 et al.! 🚀 We introduce a probabilistic inference approach for inference-time scaling of LLMs using particle-based Monte Carlo methods—achieving 4–16x better scaling on math reasoning tasks and O1-level performance on MATH500.
[1/x] can we scale small, open LMs to o1 level? Using classical probabilistic inference methods, YES! Joint @MIT_CSAIL / @RedHat AI Innovation Team work introduces a particle filtering approach to scaling inference w/o any training! check out probabilistic-inference-scal…
Akash Srivastava retweeted
🧩 Why do task vectors exist in pretrained LLMs?
Our new research uncovers how transformers form internal abstractions and the mechanisms behind in-context learning(ICL).
Akash Srivastava retweeted
Neural activity is correlated among animals performing the same task and across sequential trials.
Led by @zhang_yizi and @hl3616, we develop an reduced-rank model that exploits shared structure across animals to improve neural decoding.
biorxiv.org/content/10.1101/…
Akash Srivastava retweeted
What will a foundation model for the brain look like?
We argue that it must be able to solve a diverse set of tasks across multiple brain regions and animals.
Check out our preprint where we introduce a multi-region, multi-animal, multi-task model (MtM): arxiv.org/abs/2407.14668
Akash Srivastava retweeted
🚀 Stronger, simpler, and better! 🚀
Introducing Value Augmented Sampling (VAS) - our new algorithm for LLM alignment and personalization that outperforms existing methods!
Akash Srivastava retweeted
Excited to give a talk on our hottest, newest work “Value Augmented Sampling for Language Model Alignment and Personalization” at 2:30p Halle A3 in #ICLR2024 Reliable and Responsible Foundation Models Workshop 🥳🥳
📢Workshop on Reliable and Responsible Foundation Models will happen today (8:50am - 5:00pm). Join us at #ICLR2024 room Halle A 3 for a wonderful lineup of speakers, along with 63 amazing posters and 4 contributed talks! Schedule: iclr-r2fm.github.io/#program.
Attending #ICLR2024, interested in continual learning and like probabilistic modeling? Lazar from the @MITIBMLab, will be presenting our latest work that takes a probabilistic approach to modular continual learning on Tuesday, 7 May, Halle B #222 (iclr.cc/virtual/2024/poster/…).
I’ll be presenting our #ICLR2024 paper on a probabilistic approach to scaling modular continual learning algorithms while achieving different types of knowledge transfer. (arxiv.org/abs/2306.06545, in collaboration with @variational_i @swarat @RandomlyWalking ). A tldr (1/8):
Akash Srivastava retweeted
Check out our work titled "From Automation to Augmentation: Redefining Engineering Design and Manufacturing in the Age of NextGen-AI", where we highlight the requirements for NextGenAI suitable for design, engineering, and manufacturing.
mit-genai.pubpub.org/pub/9s6…
Instead of continuing to emphasize automation, a human-centric approach to the next generation of #AI technologies in #manufacturing could enhance workers' skills and boost productivity.
mit-genai.pubpub.org/pub/9s6…
@AustinLentsch @DAcemogluMIT @baselinescene @_faezahmed @MITMechE
