Solved WSDM Cup 2025 : Predict the better LLM response (A or B) for a prompt. Dataset was labeled by domain experts across multiple languages. Built two models—USE + LightGBM and fine-tuned DeepSeek-7B with QLoRA, handling long context via smart slicing.

a2024

WSDM Cup 2025: Evaluating LLMs with Embeddings & Fine-Tuning

A real-time AI-powered predictive maintenance system using flight telemetry, deep learning, and big data to cut downtime and optimize reliability across 200+ aircraft.

IMG 0036

Predictive Maintenance Method Based on Machine Learning for Aviation

Left to their own devices, LLMs can’t recall things outside their training set like news after training or your private docs. Augmented generation solves this by letting the model use external knowledge at inference time.

From RAG to RAG+CAG: A Field Guide to Hybrid Knowledge

When we think of Artificial Intelligence today, most people imagine huge models like GPT-4 or Claude — massive brains trained on trillions of words, capable of writing essays, generating code, or even composing songs.

But here’s the twist: most AI tasks don’t need a giant brain.

In fact, NVIDIA researchers argue that Small Language Models (SLMs) may actually be better for the jobs AI agents do most often. And that shift could change the way we design, build, and use AI.

blog1

Why Small Language Models Could Be the Future of AI Agents

OpenAI just dropped GPT-OSS — an open-weight, agent-ready model that runs locally and supports tools, code, and RAG workflows. Here’s a no-fluff guide to what it is, why it matters, and how to use it in 60 seconds.

🔓 GPT-OSS: OpenAI’s Quiet Power Move You Shouldn’t Miss

Welcome to my first blog post on this platform! My name is Mgh(Mamidipalli Ganeshhemanth ), and I’m thrilled to share a slice of my world with you. My journey in data science began five years ago, entrenched within the dynamic Retail Industry. Here, I honed my skills in predictive modelling, working closely with startups to enhance various facets of their customer interactions, from acquisition and relations to retention. This freelance endeavour has not only been my playground for practical application but also a crucible where theory meets reality.

DALL·E 2024-11-23 18.25.13 - A visually striking cover image for a Medium blog titled -Ganeshhemanth Mamidipalli’s Fusion of Data Science and Real-World Insights.- The design feat

Ganeshhemanth Mamidipalli’s Fusion of Data Science and Real-World Insights

🚀 Ready to embark on an exhilarating journey from your current domain to the captivating world of data science? Buckle up and fasten your seatbelts, as we navigate the roadmap together, filled with excitement, challenges, and a treasure trove of data! 📊💡

Roadmap for Transitioning to Data Science from Other Domains

If you’re looking to create a data science portfolio website that not only showcases your skills but also gets you noticed by employers, you’re in the right place. In this guide, I’ll show you how to build a portfolio website with ChatGPT — step by step. Best of all, you don’t need any prior coding experience, and we’ll keep it simple and actionable.

Portfolio

Build Your Data Science Portfolio Website in an Hour Using ChatGPT

Generative AI refers to algorithms that can create new content, such as text, images, or even music, by learning patterns from existing data. Unlike traditional AI, which focuses on recognizing patterns and making predictions, generative AI can produce creative outputs that mimic human abilities. These outputs can range from writing essays and poems to generating realistic images and synthesizing new music.

0 TJ26Gu0QFjRrXpl2

Power of Generative AI and Transformer Models Ganesh Mamidipalli

In the era of open large language models (LLMs), customization has become not just a luxury but a necessity. 