Amit Deshpande’s Post

8.7% of Python package hallucinations by AI models match valid JavaScript package names. Think about that. An AI suggests a Python package that happens to share a name with a real JS package. The developer doesn't find it on PyPI, gets confused, and either installs the wrong thing or an attacker registers it on PyPI first. Cross-ecosystem confusion is a real and growing attack vector.

Google ADK: Build Your First AI Agent in Python Google ADK: Build Your First AI Agent in Python Step-by-step guide to building, testing, and deploying AI agents with Google Agent Development Kit https://lnkd.in/gUwM5YDF https://lnkd.in/gUwM5YDF

Most data isn’t behind a paywall… it’s behind a login. 🔐 And that’s where many scraping workflows stop. This video shows how to scrape authenticated pages using Python, with Crawlbase supporting reliable data extraction so you can access what’s actually behind the login, not just what’s publicly visible. What you will learn: 🔹 How to handle login based scraping workflows 🔹 How session management works in real scenarios 🔹 Why authenticated scraping is critical for real data access 🔹 How Crawlbase supports more consistent data extraction If you’re working with protected data sources, this is a practical walkthrough worth watching. 👉 Watch full video: https://lnkd.in/gxfJPxGH #Crawlbase #WebScraping #Python #Automation #DataEngineering #Developers #APIs #TechTools #PythonTutorial

"BREAKING: China has open-source a massive Python framework for building AI agents called AgentScope, a python framework built around Agent-Oriented Programming that lets you build AI agents visually with MCP tools, memory, rag, and reasoning capabilities. 100% Open Source." ➡️ Helps your workflow? [https://lnkd.in/gAknPxGE)

Today I made a simple Python Debugger Agent. how it works: simply paste your python code and it will analyze Python code execution, catch errors, and suggest fixes. Scope: Debugging scripts, explaining errors, and guiding users interactively. The link is given below. You can also make it yours custom Ai agents. https://lnkd.in/d985cKZV #gensparkai #ai

Behind every working feature, there are: • 10 failed attempts • 20 Google searches • 50 lines of debug logs But that’s the process. Building scalable systems, fixing issues, and improving performance — that’s what makes development interesting. Enjoying the journey of solving problems one step at a time. #DeveloperLife #CodingJourney #Python #APIs

Rebuilt Bayesian Optimization from Python/BoTorch to native C++/CUDA. 33x faster end-to-end on a V100 — from 8.1s to 243ms per iteration. Optimizes trading strategy parameters across 19 dimensions with 20 parallel L-BFGS restarts on separate CUDA streams, all sub-second end-to-end. What made the biggest difference: - Batching cuBLAS dtrsm (one call with 4096 RHS vs per-candidate) - GPU-side perturbation generation (eliminated 95% of host-device transfers) - Shared-memory parallel reduction for MC sampling (23x kernel speedup) - Multi-stream parallel restarts (10 concurrent L-BFGS optimizations) What's your experience moving Python ML workloads to native CUDA?

Your Model Card is just as important as your Python script. Documentation should never be treated as an afterthought or a "nice-to-have" feature. A well-written Model Card explaining the intended use, data limitations, and ethical considerations is vital for transparency. It provides the necessary context for stakeholders to trust the "intelligence" you’ve built. If it isn't documented, it isn't production-ready. #ResponsibleAI #MLOps #Documentation

Built a Web Scraper with Pagination using Python. Features: 1.Scrapes quotes and authors from multiple pages 2.Implements pagination to fetch data beyond a single page 3.Allows user to control the number of results 4.Handles invalid inputs and stops when pages end This helped me understand: 1.How pagination works in web scraping 2.Loop-based data collection across pages 3.Structuring scraped data for better readability.Cognifyz Technologies