ilanr (@ilan_raman) profile picture

Research and Development @ Interface Masters Technologies.

Joined October 2012
Photo by ilanr (@ilan_raman)Photo by ilanr (@ilan_raman)Photo by ilanr (@ilan_raman)Photo by ilanr (@ilan_raman)Photo by ilanr (@ilan_raman)
Microsoft just put a tool on GitHub that makes your coding agent cheaper. It doesn't touch the agent. TokenEconomizer learns from the sessions your agent already ran in a repo, then cuts the waste in the next ones: → It remembers repo facts your agent keeps rediscovering, like the build step that has to run first → It stops loading tools your agent never uses → On SWE-Bench: 18% to 31% lower cost across 5 agent and model combos, with accuracy up in 4 of 5 Built by Microsoft Research and UC Berkeley. github.com/microsoft/TokenEc…
28
21
3
213
56,839
ilanr retweeted
Microsoft just made Claude Code and Codex cheaper without touching the agent. TokenEconomizer is a free, open-source plugin that learns from the sessions your agent already ran in a repo: - Remembers the repo facts your agent keeps rediscovering - Stops loading tools that never get used, since every tool definition is sent with every request - Works with Claude Code, Codex, Copilot CLI and Pi - 18.4% to 30.9% lower cost across 5 agent-model setups on SWE-Bench If you hit your usage limits every week, this is free money. Save your tokens. github.com/microsoft/TokenEc…
55
200
12
1,765
108,467
Must for Musk fans. silicon-valley.tech
37
97
3
897
137,367
Bring your ideas to life with Zeus, a new type of GPU: - Expandable memory: 32/64/128 GB soldered + 2x/4x SO-DIMM slots for up to 384 GB memory! - Massive improvements to path tracing performance! - 400 GbE QSFP-DD port! - Built-in high-performance RISC-V CPU cores capable of running Linux! - DevKits 2026, Mass Production 2027 And we used the good old 8-pin PCIe power connector that is known to not melt!
294
572
181
5,608
27,913,361
LLM visualization (by Brendan Bycroft): bbycroft.net/llm
17
127
9,002
Introducing sdplane, a DPDK-based router: mails.dpdk.org/archives/user… We are developing another router implementation using DPDK, named "sdplane" (for software data plane). We released a new version under the GPLv3 yesterday, and we think you might find it interesting. github.com/kait-cronos/sdpla… We aim for a full-fledged and yet extensible router. It is not complete yet, but full multicast and SRv6 routing is planned to be working by March 2027. It uses RCU (liburcu), is compact (around 40K lines of C, excluding bundled code), and, compared with VPP, focuses on multicast. It also integrates Pktgen-DPDK as a traffic generator. Debian packages for Ubuntu 24.04 and 26.04 are available on the releases page. We would appreciate it if you could test them. github.com/kait-cronos/sdpla… At the Supercomputing (SC26) conference next week, our university (KAIT) will give a demonstration using sdplane: prlog.org/13180316-kait-to-d… Our project is supported by JST CRONOS (Grant No. JPMJCS24N9).
18
28
1,491
openai's head of design's website omg iansilber.com/ 😭
571
655
454
13,521
5,848,372
Son Dakika. Robotik öğrenmenin en pahalı kısmı ortadan kalktı. Bir geliştirici, Fable 5.2 ile tamamen interaktif bir “Robotik Mühendisliği” eğitim sitesi açtı. Artık GPIO pinini öğrenmek için 500$+ başlangıç kiti almana gerek yok: • 3D ortamda 8 gerçek kart: Arduino Uno, Raspberry Pi 4 ve diğerleri • Hiper gerçekçi grafikler, gerçek hayattaki hareketlerle birebir • Parça parça inşa et: Çipleri, konektörleri ve başlıkları yerleştirip lehimleyebiliyorsun • Gerçek devreleri bağla: LED yak, sensör oku, motor çalıştır En dikkat çekici detay: Tüm 3D modelleri Fable 5.2 kodla üretti. Hazır 3D dosya veya fotoğraf kullanılmamış. Ücretsiz ve kayıt gerektirmiyor. robotics-lab-app.vercel.app/… Bu tarz projeler çoğalırsa donanım eğitimi tamamen değişir.
11
525
15
3,245
179,777
this is crazy.. ngl firstprinciplesapproach.dev
12
74
1
736
64,922
Announcing two open-source projects for building and operating managed Kubernetes services on GPU infrastructure: NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR). Both are components of the NVIDIA DSX AI Factory Platform. NVKE is the control plane for running managed Kubernetes service and NVKR builds the OS images, container images, and packages the managed clusters depend on. Available in public preview, get started: nvda.ws/4zCSpMp
25
133
17
982
76,989
We just killed Slack, and open-sourced it. Introducing Terminus Teams: we kept everything you like about Slack (channels, calls and meetings, threads) but kicked out the Slack bot, so you can use your existing Claude Code or Codex to read messages and even reply. 100% free, 100% open source, 100% customizable. Export your data anytime.
95
72
22
731
128,313
Updated Protocol Explorer with more protocols! New: VPNs (WireGuard, IPsec), WebSockets, WebRTC, SMB, BGP, IPv6, Wi-Fi & WPA3, DNS over HTTPS, proxies, and more. Plus search, keyboard shortcuts, a diagram-only view, and support for 21 languages. protocol-explorer-alpha.verc…
51
200
16
1,579
77,944
NVIDIA Open-Sources Auto Ontology, a Semantic Layer for AI Agents Your boss asks the AI agent: "What was revenue last quarter?" It answers in 3 seconds. Clean number. Nice chart. But it used the wrong "revenue" column. 𝗡𝗼𝗯𝗼𝗱𝘆 𝗻𝗼𝘁𝗶𝗰𝗲𝘀. That's the problem NVIDIA's new open-source library, Auto Ontology, goes after. Its README puts it plainly: "…𝘢 𝘳𝘦𝘷𝘦𝘯𝘶𝘦 𝘮𝘦𝘵𝘳𝘪𝘤 𝘢𝘯𝘴𝘸𝘦𝘳𝘦𝘥 𝘧𝘢𝘴𝘵𝘦𝘳 𝘪𝘴 𝘸𝘰𝘳𝘵𝘩 𝘯𝘰𝘵𝘩𝘪𝘯𝘨 𝘪𝘧 𝘵𝘩𝘦 𝘯𝘶𝘮𝘣𝘦𝘳 𝘪𝘴 𝘸𝘳𝘰𝘯𝘨." 𝗪𝗵𝗮𝘁 𝗶𝘁 𝗶𝘀: A "context layer" that sits between an AI agent and your database. It tells the agent where the data lives, what each field means, and whether to use it. 𝗛𝗼𝘄 𝗶𝘁 𝘄𝗼𝗿𝗸𝘀, 𝘀𝘁𝗲𝗽 𝗯𝘆 𝘀𝘁𝗲𝗽: • 𝗗𝗶𝘀𝗰𝗼𝘃𝗲𝗿: it reads your database metadata and drafts an ontology, a map of business terms like "active customer" • 𝗚𝗿𝗼𝘂𝗻𝗱: agents use those terms instead of guessing from raw table names • 𝗩𝗮𝗹𝗶𝗱𝗮𝘁𝗲: it checks generated SQL against certified definitions • 𝗘𝘅𝗽𝗼𝘀𝗲: agents reach it through MCP, a CLI or an API • 𝗟𝗲𝗮𝗿𝗻: certified corrections go back into the map 𝗙𝗼𝗿 𝘂𝘀𝗲𝗿𝘀: Ask a question, get 𝘁𝗵𝗲 𝗮𝗻𝘀𝘄𝗲𝗿, 𝘁𝗵𝗲 𝗦𝗤𝗟 𝗶𝘁 𝗿𝗮𝗻, 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗿𝗼𝘄𝘀. You can check the work, not just trust it. 𝗙𝗼𝗿 𝗯𝘂𝗶𝗹𝗱𝗲𝗿𝘀: • Apache 2.0, runs with Docker Compose or Helm • Works with Databricks, Postgres, Snowflake and DuckDB • Defaults to NVIDIA's Nemotron models, but 𝗮𝗻𝘆 𝗢𝗽𝗲𝗻𝗔𝗜-𝗰𝗼𝗺𝗽𝗮𝘁𝗶𝗯𝗹𝗲 𝗲𝗻𝗱𝗽𝗼𝗶𝗻𝘁 𝘄𝗼𝗿𝗸𝘀 • The MCP server has 𝟵 𝗿𝗲𝗮𝗱-𝗼𝗻𝗹𝘆 𝘁𝗼𝗼𝗹𝘀. There is deliberately no tool to browse raw tables, so agents must go through the business terms 𝗜𝘁 𝗯𝘂𝗶𝗹𝗱𝘀 𝗼𝗻 𝗲𝗮𝗿𝗹𝗶𝗲𝗿 𝘄𝗼𝗿𝗸: • Cube and dbt MetricFlow: semantic layers you write by hand in code • Wren AI: an open semantic layer for AI agents, with a context layer built partly by agents • Snowflake's Semantic View Autopilot: also drafts semantic models with AI, but only for Snowflake What's different here: 𝗼𝗽𝗲𝗻 𝘀𝗼𝘂𝗿𝗰𝗲, 𝘄𝗼𝗿𝗸𝘀 𝗮𝗰𝗿𝗼𝘀𝘀 𝗳𝗼𝘂𝗿 𝗱𝗮𝘁𝗮𝗯𝗮𝘀𝗲𝘀, 𝗮𝗻𝗱 𝗯𝘂𝗶𝗹𝘁 𝗳𝗼𝗿 𝗮𝗴𝗲𝗻𝘁𝘀 𝘁𝗵𝗿𝗼𝘂𝗴𝗵 𝗠𝗖𝗣. 𝗧𝗵𝗲 𝗵𝗼𝗻𝗲𝘀𝘁 𝗽𝗮𝗿𝘁: • 𝗧𝗵𝗲 "𝟵𝟬% 𝗳𝗮𝘀𝘁𝗲𝗿" 𝗰𝗹𝗮𝗶𝗺 𝗶𝘀 𝗡𝗩𝗜𝗗𝗜𝗔'𝘀 𝗼𝘄𝗻 𝘁𝗲𝘀𝘁𝗶𝗻𝗴, with no baseline published • 𝗡𝗼 𝗮𝗰𝗰𝘂𝗿𝗮𝗰𝘆 𝘀𝗰𝗼𝗿𝗲𝘀 𝗮𝗿𝗲 𝗽𝘂𝗯𝗹𝗶𝘀𝗵𝗲𝗱. NVIDIA ships an eval harness (BIRD, 1,534 questions) but no results • That harness was tested with 𝗖𝗹𝗮𝘂𝗱𝗲 𝗢𝗽𝘂𝘀 𝟰.𝟴, 𝗻𝗼𝘁 𝗡𝗲𝗺𝗼𝘁𝗿𝗼𝗻 • 𝗪𝗶𝘁𝗵𝗼𝘂𝘁 𝗛𝗮𝘀𝗵𝗶𝗖𝗼𝗿𝗽 𝗩𝗮𝘂𝗹𝘁, 𝗱𝗮𝘁𝗮𝗯𝗮𝘀𝗲 𝗽𝗮𝘀𝘀𝘄𝗼𝗿𝗱𝘀 𝗮𝗿𝗲 𝘀𝘁𝗼𝗿𝗲𝗱 𝘂𝗻𝗲𝗻𝗰𝗿𝘆𝗽𝘁𝗲𝗱 in Postgres (per the README) • It's young: 𝗮𝗯𝗼𝘂𝘁 𝟳𝟱𝟬 𝗚𝗶𝘁𝗛𝘂𝗯 𝘀𝘁𝗮𝗿𝘀, and NVIDIA isn't accepting outside contributions 𝗧𝗵𝗲 𝗯𝗶𝗴𝗴𝗲𝗿 𝗽𝗼𝗶𝗻𝘁: the hard part of AI analytics isn't writing SQL. 𝗜𝘁'𝘀 𝗮𝗴𝗿𝗲𝗲𝗶𝗻𝗴 𝗼𝗻 𝘄𝗵𝗮𝘁 𝘁𝗵𝗲 𝘄𝗼𝗿𝗱𝘀 𝗺𝗲𝗮𝗻. GitHub: github.com/NVIDIA/auto-ontol… Links in reply 👇
109
323
36
3,116
342,343
Linux Kernel Internals. kernel-internals.org/
2
55
592
26,522
Ever nodded along when someone mentioned TCP handshakes or DNS lookups, without really knowing how they work? Then this is for you. A visual quick reference for 14 internet protocols, with a built-in packet inspector. protocol-explorer-alpha.verc…
70
1,472
157
10,296
959,068