Software Engineer / Year IV
SAMIUL HOSSAIN
I build systems: kernel schedulers, real-time game engines, full-stack platforms. My main focus is local AI, a fully functioning assistant that runs on hardware I own.
The short version
I'm a fourth-year software engineering student who learns by building things that have to actually work.
My education runs on two tracks. The classroom gave me the theory: operating systems, software design, concurrency. The depth came from the projects I chose for myself, like rewriting a CPU scheduler inside a real kernel, architecting a strategy game with two interlocking simulation layers, and shipping a scheduling platform that a real community depends on.
Right now most of my energy goes into local AI. I'm building a functioning local assistant that runs entirely on my own hardware: open-weight models, quantized inference, retrieval, and the systems glue that makes it fast enough to feel alive. No cloud, no API keys, full control.
That range is deliberate. Going from C and RISC-V context switches up to TypeScript APIs and model runtimes means I understand the whole stack under the software I write, and I'm comfortable being dropped anywhere in it.
Cinderfall
- Godot 4
- GDScript
- Real-time Simulation
A grand strategy game with real-time tactical combat. Campaign layer, unit AI, and combat resolution built from scratch, with two interlocking simulation modes running inside a single engine loop.
"One wrong assumption about save state and the whole campaign desyncs. Designing for that taught me more than any lecture."
open case studyttgod-scheduler
- TypeScript
- REST API
- Discord Bot
A squad scheduling platform: availability matrices, campaign planning, and a Discord bot API surface. End-to-end ownership from data model to deployed interface, built for a real community.
"The hard part wasn't the code. It was availability resolution across timezones, partial overlap, and recurring sessions."
open case studyWhat I work with
Local AI & Inference
The main quest: a fully functioning assistant that never leaves my machine.
- Open-weight LLMs (Llama, Mistral)
- llama.cpp / Ollama runtimes
- Quantization & GGUF
- RAG & local embeddings
- Tool-use & agent pipelines
Systems & Low-Level
Close to the machine, where every decision has a cost.
- C / RISC-V assembly
- xv6 kernel internals
- CPU scheduling & context switching
- Concurrency & synchronization
- Memory models
Web & Platforms
End-to-end product engineering, data model to deployed interface.
- TypeScript / Node.js
- REST API design
- Discord bot integrations
- Relational data modeling
- Git, CI, code review
Simulation & Games
Real-time loops, interlocking state machines, and worlds that tick.
- Godot 4 / GDScript
- Game architecture & ECS thinking
- Deterministic combat systems
- Python tooling & TUIs
- Event-driven design
Currently: finishing my degree, teaching my local AI to be actually useful, and looking for a team that ships hard things.