Log inSign up
Log inSign up
Deepu K Sasidharan
11.8K posts
Deepu K Sasidharan profile banner
@deepu105

Deepu K Sasidharan

@deepu105
@[email protected] 🍉 🇵🇸 @jhipster co-chair. Developer 🥑 @okta. Polyglot dev/Speaker/Author. @Java_Champions. Java, Rust, JS, DevOps. ADHD.
The Netherlands
deepu.tech
Joined November 2009
1,174 Following
4,617 Followers
RepliesRepliesRepostsRepostsMediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·US TIDA·Ads Info·© 2026 X Corp.
  • Pinned
    @deepu105
    Deepu K Sasidharan
    @deepu105
    Jun 2
    Today I'm releasing LlamaStash 0.0.2: a zero-overhead, terminal-native launcher for llama.cpp. One Rust binary that's a TUI, a CLI, a daemon, and an OpenAI-compatible proxy. Demo below 🧵
    GIF
    2
  • @deepu105
    Deepu K Sasidharan
    @deepu105
    Oct 8
    Just published pi-automode-classifier, an auto mode plugin for the Pi coding agent that uses Jev or Kev/Laya (running locally) to classify commands. Risky commands need my approval before they run. pi install npm:pi-automode-classifier
    GitHub - deepu105/pi-automode-classifier: Auto mode for Pi: rules first, then a local decision...
    From github.com
    3
  • @deepu105
    Deepu K Sasidharan
    @deepu105
    Oct 6
    I gave the same complex Rust feature to Opus 5.5 and Qwen3.8-Flash-Next running locally. Opus did it 7 times faster but Qwen made the better PR which eventually got merged Full writeup
    Forget Claude Code. All you need is Qwen3.8-Flash-Next running locally for agentic coding |...
    From deepu.tech
  • @deepu105
    Deepu K Sasidharan
    @deepu105
    Oct 2
    LlamaStash v0.6.0 is out 🦙 Highlights: - a request that doesn't fit now unloads idle models to make room, and 'presets save --idle-ttl 0 --preload' keeps a model warm. - Claude Code's /effort now reaches llama.cpp models.
    LlamaStash — terminal-native local-LLM manager (TUI + CLI)
    From llamastash.dev
  • @deepu105
    Deepu K Sasidharan
    @deepu105
    Sep 30
    I benchmarked engines for Qwen 3.8-Flash-Next on Strix Halo (Flow Z13, 128 GB). Halogen with its native weights is fastest at 39-40 t/s decode, then gufo and CIRU. But Halogen is closed source. gufo is open source and loads 4x faster from cold.
    From the LocalLLM community on Reddit: Benchmarks: Best engine for Qwen 3.8-Flash-Next on Strix Halo
    From reddit.com