Nandini Lokesh Reddy·May 16So anything we can measure, we can optimise: Verifiers in AIWhy do AI/LLMs generate some answers very fast & some very slow?
Nandini Lokesh Reddy·Feb 5Deepseek OCR- 2For a long time, progress in vision–language models has been narrated through benchmarks: higher accuracy, lower edit distance, better…
Nandini Lokesh Reddy·Jan 7Dolphin by BytedanceAs the year comes to a close, I ended up reading a lot of work on document parsing and image parsing, from LLM makers as well as enterprise…
Nandini Lokesh Reddy·Dec 3, 2025LangChain <> LangGraph: Chains and CyclesI was recently discussing the tech stack of our AI agent environment with a friend. I mentioned that we use both LangGraph and LangChain…
Nandini Lokesh Reddy·Nov 22, 2025Qwen-3Large Language Models have been growing in two directions for years: more data and more parameters. But the real question we keep coming…
Nandini Lokesh Reddy·Nov 14, 2025DeepSeek OCRWhen we say “a picture speaks a thousand words,” it actually makes sense in real world.