Building open-source tools and independent software for people who create.
Systems programming · Developer experience · Local-first applications · Creative tools · Learning in public
🌐 Website · 💻 Projects · ✍️ Blog · 💬 Contact
I'm a developer interested in what happens under the hood: from Rust, C++, and Assembly to cross-platform desktop apps, TypeScript interfaces, and educational software.
Most of my projects are open source, so people can inspect, understand, test, and improve them. I also develop select products privately. Across both, I value privacy-conscious defaults, clear documentation, useful automation, reproducible builds, and release notes that distinguish finished features from experiments.
| Project | What I'm building | Explore |
|---|---|---|
| 🧬 RepoDNA | Local-first codebase intelligence: architecture, repository history, dependencies, and evidence-backed reports; also has a browser version. | Web version ↗ |
| ✨ LineLoom | VS Code inline completions powered by your own local Ollama model, with user-controlled behavior. | VS Code Marketplace ↗ · Open VSX ↗ |
| 🎬 FrameTrellis | Privacy-focused multi-track desktop video editing with C++20, Qt/QML, MLT, and FFmpeg. | Downloads & releases ↗ |
| 🧩 ReproCrate | Rust-based diagnostics, bounded log analysis, privacy-reviewed reports, and portable reproduction bundles. | Experimental releases ↗ |
| 📚 LangSteps | Interactive programming education across ten learning paths, with lessons, examples, quizzes, and progress tracking. | Explore lessons ↗ |
| ⚙️ Rewind64 | Experimental Linux x86-64 debugger with Assembly-written inspection logic and instruction stepping. | Source & pre-releases ↗ |
Arvune is my proprietary, closed-source, local-first AI assistant being developed for personal productivity and software creation. The goal is to bring conversations, coding workflows, and on-device inference into one user-controlled experience, without requiring a cloud AI account.
What I'm building:
- 💬 Private AI chat: Local conversations, chat search, editing and regeneration, history management, and user-initiated backups.
- 💻 AI coding workspace: Work with explicitly approved local project files, review suggested changes before applying them, and use checkpoints and rollback. Imported content does not trigger automatic command execution.
- 🧠 Original local AI models: An in-house Rust inference runtime and model experiments, with explicit model imports and no required remote inference service.
- 🛡️ Privacy and control: Local data storage, optional explicit exports, and no required account or application telemetry.
- 🖥️ Multi-platform ambition: Windows-first desktop development, Linux/macOS build targets, and experimental Android and browser experiences.
Release status: Arvune is in private development, with a public product release planned. A stable public download and launch date have not been announced. Model quality, performance, installation, and device-specific testing remain important release gates. Descriptions above reflect current development and planned direction, not a promise that every feature will be included at launch.
Source availability: Arvune's GitHub repository is private and its source code is not open source. Public product information and launch updates will be shared through my website when available.
Some of my other work is intentionally experimental. A repository or pre-release does not imply a production-ready app.
| Project | Current direction |
|---|---|
| 🎙️ KathaSvara · कथास्वर | Offline-first Hindi storytelling and speech-production desktop tooling; Hindi narration quality and standalone packaging still need validation. |
| 🔗 PairDock | Connected-device workspace spanning firmware, API, SDK, and dashboards; currently a development foundation, not a production device service. |
| 📄 foliolane | Local-first document-merging tools for desktop, CLI, and browser-hosted workflows; repository documentation is being transitioned from its former DocMergeForge name. |
| Area | Projects |
|---|---|
| 🖥️ Desktop & productivity | EyeZen · CalcNova · Markora |
| 🌐 Web & full-stack | SpeedPulse (web app) · CampusCore |
| 🎮 Games | SudokuNova · Latebrake (Unreal Engine source preview; not a playable release) |
| 📖 Learning resources | Learning Rust · The Journey of Ethical Hacking |
| Focus | Tools and languages |
|---|---|
| Systems & performance | Rust · C · C++20 · x86-64 Assembly · NASM |
| Web & services | TypeScript · JavaScript · React · Node.js · ASP.NET Core |
| Desktop & mobile | Kotlin · Jetpack Compose · Flutter · Qt/QML · Tauri · C#/.NET · Avalonia |
| Tooling & testing | Python · Git · GitHub Actions · CMake · CI · Automated tests · Windows · Linux |
These technologies are used across different projects; individual repositories describe their actual requirements.
- Show the evidence: Document implemented behavior, tests, limitations, and incomplete acceptance checks.
- Keep users in control: Prefer local processing, explicit sharing, and privacy-aware defaults where appropriate.
- Make contribution easier: Invest in setup instructions, issue tracking, readable source code, and reproducible workflows.
- Keep experimenting: Learn through systems software, developer tooling, educational products, and creative applications.
I'm open to constructive issues, pull requests, code reviews, documentation help, testing, and project collaborations. For a specific open-source tool, the best first step is its repository's Issues or Discussions section (where enabled). For private products such as Arvune, use my public contact page.
Find me: Website · Blog · LinkedIn · X · dev.to · YouTube · Instagram · Bluesky
Project and business enquiries: Contact me through my website.
Support independent software and open-source development (optional): Buy Me a Coffee · Razorpay · PayPal
Build → Learn → Improve → Share.