Mobile Engineer Flutter & Dart || Bitcoin and LIghting || Opensource Contributor.

Abuja_Nigeria
Joined November 2020
Hello everyone 👋 My name is Usman Umar Garba I’m a 3rd-year Computer Engineering student and a Mobile Engineer (Flutter), actively building cross-platform applications. I’m currently expanding into backend development with Python (APIs) and exploring Bitcoin &
4
15
522
Code Warrior retweeted
How I stay consistent with my 365 Days of Code challenge. A lot of people have been asking me how I manage to show up every day, write code, build projects, and share updates everyday. Here’s what has helped me: 1. I don’t wait for motivation. Some days I’m excited to code. Other days, I just want to rest. I’ve learned not to depend on how I feel. I show up regardless. 2. I work on real projects. I’m not just coding for the sake of completing the challenge. I work on products, experiment with new technologies, fix bugs, and build features. Having something meaningful to work on makes it easier to keep going. 3. I document my progress. Sharing what I’m building publicly gives me a way to track my growth. When I look back at older posts, I can see how far I’ve come, and that motivates me to keep going. 4. I don’t try to do everything in one day. I don’t just wake up and start coding randomly. I create a plan for what I want to build and break it down into features I can implement step by step. This gives me direction and helps me stay focused. 5. I focus on showing up, not being perfect. There are days when things don’t go as planned. I’ve learned that consistency isn’t about having a perfect routine; it’s about finding a way to keep moving forward. I’m currently on this journey, and one thing I’ve learned is that consistency becomes easier when you build a system around it instead of relying on motivation alone. You don’t have to do something extraordinary every day. Just make sure you’re moving forward. If you’re working on a long-term goal, start today. Document your progress, stay accountable, and keep going. I’m currently building Coddle Learn, an open source platform for developers and techies to learn, practice and build their portfolio.
15
12
3
166
2,141
Code Warrior retweeted
What goes into building a stronger Bitcoin ecosystem, beyond the code itself? Our conversations with @StephTitcombe, @Murugi___ and @ihate1999 took us behind different parts of Btrust's work, from creating pathways for developers to grow, to communicating the work happening across the ecosystem and the vision guiding it all. Each conversation offered a different perspective on what it takes to support the people building Bitcoin's future.
4
18
51
1,150
Code Warrior retweeted
Big congratulations to @dev_useee and @Muhamma90584326 for emerging as winners of the Machine Money Track at Boss Battle 2026! 🎉👏 Their project, INFERNOS, showcases the power of permissionless AI inference powered by Bitcoin Lightning. ⚡₿ Congratulations to the winners
1
7
69
🇳🇬🏆 We won the Machine Money Track at BOSS Battle 2026! Huge thanks to @bitshala_org, @btrust_builders the organizers, judges, and everyone who supported us. We built Infernos — permissionless AI inference paid with Bitcoin Lightning. Here's the story behind our journey 👇 📝
Replying to @bitshala_org
Next up, the Machine Money Track: Bitcoin + AI 🏆 The winner is Infernos by @dev_useee & team from Nigeria 🇳🇬 Permission-less AI inference, paid with Lightning. Imagine being able to access AI compute and pay for it in sats, without needing anyone’s permission. That’s what makes projects like this exciting. Congratulations to the Team! 🧡 🔗 github.com/infernos-ai/infer…
8
6
2
18
400
Replying to @AdewumiSaheedA2
@AdewumiSaheedA2 showed how agentic apps can transform manual invoice workflows with autonomous agents, powered by ADK, MCP, and AG-UI. #DevFestAdoEkiti #DevFest2026 #GDGAdoEkiti
3
11
277
We fixed it by making the mock Lightning flow behave much more like the real protocol: Invoice → payment hash → preimage → SHA-256 verification → settlement → L402 authorization That forced us to separate: Payment from Proof from Authorization.
1
3
40
And the project is open source. You can inspect the implementation, architecture, tests and development history here: github.com/infernos-ai/infer… We'd love people to read it, break it, review it and tell us what we got wrong.
1
3
32
Can open AI compute become a permissionless economic resource? We don't think we've solved the whole problem. But we built a working foundation for exploring it. Permissionless AI Compute. Pay in Sats. Earn from Compute. ⚡🤖🇳🇬 #Bitcoin #Lightning #AI @bitshala_org @tech_lateef
4
32
What if AI compute could become a permissionless service that machines can pay for directly with Bitcoin? Over the last few weeks, we tried to build exactly that. Introducing Infernos. Permissionless AI Compute. Pay in Sats. Earn from Compute. ⚡🤖 🧵👇
Building Infernos 🔥 permissionless open-model inference paid in sats. We’ve now got the core flow working: Request → 402 → Pay → Prove → Unlock → Infer ⚡ Rust/Axum node 🤖 Ollama open models 🔐 L402 + capability-gated authorization 💰 Session budgets + dynamic pricing
2
6
115
At one point our CLI could receive an authorization challenge, but the server rejected the payment proof. The problem? Our mock payment flow wasn't modelling the cryptographic relationship between the invoice and preimage correctly.
1
2
12
SDK + CLI We also built the Lightning backend around an abstraction so different Lightning environments can be integrated without changing the core inference gateway. The interesting part wasn't returning a 402. The hard part was making the payment actually mean something.
2
3
24
The payment proof unlocks a bounded capability. Under the hood, Infernos is built with: • Rust • Axum + Tokio • Lightning/L402 • Open-weight models • Ollama • OpenAI-compatible APIs • Session budgets • Capability-based authorization • Next.js playground • Rust caller
1
2
27
The caller pays only when it wants to use the compute. Request → HTTP 402 → Lightning payment → Payment proof → Capability verification → Inference We use L402 to connect the payment to authorization. Payment isn't just a billing event.
1
2
23
Infernos turns a machine running an open-weight model into a Lightning-paid inference endpoint. The basic relationship is: Provider → Compute Caller → Sats Infernos → Payment-gated access The provider supplies compute and earns sats.
1
2
23
We started with a simple question: How can an independent machine running an open-weight model become a service that anyone — including an AI agent — can pay for without creating an account? Most AI infrastructure today is built around centralized providers, API keys, accounts.
1
2
57