Sam Palani

🚀 My Java learning journey has officially paid off! After weeks of diving deep into a new language, I’m excited to share the results: I just published a full article on building generative AI applications in Java using LangChain4j. Java has always been a powerhouse in enterprise systems, but seeing how quickly generative AI tooling is evolving around it has been eye-opening. From structured output to RAG, I walked through how to build real GenAI workflows entirely in Java — and reflected on how it compares to the Python ecosystem we’re all so familiar with. Whether you come from the Java world stepping into GenAI, or from Python wondering how Java is adapting, I hope this deep dive gives you a clear, grounded view of where the ecosystem is heading.

9

Great article by Ashpreet B. on adaptive AI agents and continuous learning AI. Highly recommended reading for agentic engineering. https://lnkd.in/gZQknWrn

16

One way to get better results with LLMs is to grow the training corpus and build larger architectures. Another way that is emerging is to allocate more computation time for inference. In the following paper, the authors propose a novel inference time scaling approach by adapting particle-based Monte Carlo methods -- I really like that idea; partly because it intends to utilize probabilistic modelling for inference, partly because I favor SSMs, and partly because this kind of approach will make it possible to run LLMs locally with decent results. Here is the link: https://lnkd.in/dFUQnT3X Enjoy! #AI #LLM #Inference #ProbabilisticModeling

18

1 Comment

Fine-tuning LLMs with evolution strategies (an alternative to gradient descent) has turned out to be a productive research direction! Four new papers from Cognizant AI Lab evaluate it in several new tasks including Math and ARC-AGI, extend it to fine-tuning quantized models, demonstrate how LLMs can be trained to know what they know, and show why high dimensionality of models is a blessing (not a curse) in fine-tuning. Overview blog: https://cgnz.at/6005QZNMb Papers: https://cgnz.at/6007QZN13 (tasks)     https://cgnz.at/6003QZNHN (quantized)     https://cgnz.at/6002QZNGG (metacognition)     https://cgnz.at/6000QZNyI (blessing) Animated illustration: https://lnkd.in/gUUaYVAT

149

3 Comments

Autonomous Code Evolution Meets NP-Completeness Authors Cunxi Yu Rongjian Liang CHIA-TUNG (Mark) HO Haoxing Ren NVIDIA AI Abstract Large language models (LLMs) have recently shown strong coding abilities, enabling not only static code generation but also iterative code self-evolving through agentic frameworks. Recently, AlphaEvolve \cite{novikov2025alphaevolve} demonstrated that LLM-based coding agents can autonomously improve algorithms and surpass human experts, with scopes limited to isolated kernels spanning hundreds of lines of code. Inspired by AlphaEvolve, we present SATLUTION, the first framework to extend LLM-based code evolution to the full repository scale, encompassing hundreds of files and tens of thousands of lines of C/C++ code. Targeting Boolean Satisfiability (SAT), the canonical NP-complete problem and a cornerstone of both theory and applications. SATLUTION orchestrates LLM agents to directly evolve solver repositories under strict correctness guarantees and distributed runtime feedback, while simultaneously self-evolving its own evolution policies and rules. Starting from SAT Competition 2024 codebases and benchmark, SATLUTION evolved solvers that decisively outperformed the human-designed winners of the SAT Competition 2025, and also surpassed both 2024 and 2025 champions on the 2024 benchmarks. #ai #code #solvers #satlution H/T Cecile G Tamura Link to paper: https://lnkd.in/dHtNUJCQ Kelly Cohen Zack Huhn Summer Crenshaw Lohith Srikanth Pentapalli Scott Phillips

6

1 Comment

Compressing LLMs but worried about accuracy? 🎯 EoRA uses eigenspace low-rank approximation to compensate for errors—no retraining needed. A promising direction for scalable, task-adaptive LLMs. New paper from #NVIDIAResearch.

1

Exciting to see Microsoft and Google come together to formalize the Agent2Agent (A2A) protocol — a major step toward realizing the agentic web! With A2A and MCP support integrated into Azure AI Foundry and Copilot Studio, we’re entering a new era of intelligent agent collaboration. This will unlock seamless interoperability between agents and services, enabling more complex, autonomous workflows that work on behalf of users across platforms. Kudos to the teams driving this innovation! Looking forward to what’s next in the evolution of the agentic economy. #AI #AgenticWeb #A2A #Microsoft #CopilotStudio #AzureAI #Interoperability #FutureOfWork

17

Andrew Ng reported from Davos, that AI deployment for transformative impact will require workflow redesign. It is "theory of constraints" all over again (remember Eli Goldratt's "The Goal"?), we have to understand that real business benefits come only from working across silos, from analyzing flow bottlenecks, from understanding that optimizing non-bottlenecks creates waste. As Ng noted, letting a thousand (AI) flowers bloom has failed to lead to significant payoffs. Obviously! We in the silicon industry are no strangers to this. Workflows have been stagnant for decades. Technologies for high level specification, high level synthesis, and formal verification are not yet integrated into the full silicon development flow. They are often done piece-meal, and enthusiasts' pilots are dropped at the hint of a failure or delay. People say they want formal verification in their flow, but want it yesterday and do not have the time to invest in training engineers for its rigorous usage. Ashish Darbari, responding to a question on "if formal verification is so powerful then how come it is not widely used", noted that the problem is in the name - the actual task to be done is formal analysis while the bug detection is only a corollary, but the industry puts the cart before the horse. Based on my experience with industrial practice of high level specifications and formal verification, I tend to agree with his succinct comment. Jan Rabaey, in a DAC keynote speech, said that someday he expects to see a comic book series entitled "adventures in formal verification". AI deployment faces the same problem, if not actively worked on by business leaders. To be fair, deployments of all disruptive technologies face this mental barrier, we are so used to thinking in terms of the old flow (using the old technology) that thinking of a new flow literally means us having to begin again from first principles - few leaders have the courage and the time to do that. Goldratt mentioned that he himself had faced the same mindset problem when deploying the early ERP solutions and was surprised why he wasn't getting the expected large benefits, and he said it took him many years before he could identify the core problem and define solutions for it, which he named (rather academically) as "theory of constraints". AI leaders would do well to rework the conventional workflows in a holistic manner in order to get the real benefits from this new technology. As Henry Ford had said, "If you always do what you've always done, you'll always get what you've always got".

21