

Raunak
1.7K posts

@raunakdoesdev
I’m into open science, elegant software architecture, and well-crafted interfaces. Building https://t.co/0GX1wFmPoU



🔍 Introducing BackSearch. LLMs are increasingly asked to predict the future, but a good backtest requires a snapshot of the internet at a point in time. BackSearch allows LLMs to search the web as it was on a particular date. It’s great for: 🔮 Forecasting and prediction markets. 📈 Quantitative finance. 🌍 RL environments that simulate the world. 🥶 Freezing websearch for benchmark reproducibility We’re releasing a narrow slice of our index to begin with, focused on the news domain for 2026. Based on feedback we’ll open up more of our index in subsequent releases. 👇 Try BackSearch out with the link below.






"people believe OCR is a solved problem. i very much don't think so. as long as humans are here, unstructured data is here to stay." i caught up with @sidpagariya, founding engineer at @reductoai, at our @aiDotEngineer booth. @reductoai parses hundreds of millions of pages a week and turns messy, unstructured documents into structured data for healthcare, legal, and fintech teams. sid walks me through how @reductoai's in-house ML team trains its own models and now outperforms frontier models on the longest, messiest documents. we also got into: 1:51 the "holy triangle" of document parsing: accuracy, latency, and cost 3:50 building agent experience so claude or codex can one-shot "use Reducto to parse this" 5:48 training models in-house with a fast-growing ML team, hundreds of millions of pages a week 10:13 the messiest docs they've seen: 100,000-page legal packets, and why OCR still isn't solved watch the full episode here:

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter. AI will transform every industry, power every company, and be built by every country. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. The world needs both frontier closed models and frontier open models. images.nvidia.com/pdf/Open-Weigh…










We're partnering with @huggingface to investigate an unprecedented security incident. Cyber-capable OpenAI models compromised Hugging Face production during a benchmark evaluation. Sharing preliminary findings to help defenders understand emerging risks: openai.com/index/hugging-…



@ceefryingpan I think they haven't quite nailed the persistent memory piece across threads and knowing when to proactively message vs. not. I think to do this right requires using cheaper models for things like triage and route to the right context dynamically. Definitely a hard problem!


@ceefryingpan I think they haven't quite nailed the persistent memory piece across threads and knowing when to proactively message vs. not. I think to do this right requires using cheaper models for things like triage and route to the right context dynamically. Definitely a hard problem!