> Hardware-Accelerated Inference | Local LLMs (llama.cpp / Vulkan) | Deterministic Agent Systems
System Specs & Kernel Context
talha@infer-workstation
-----------------------
Role: AI Systems & Inference Engineer
Focus: Vulkan/llama.cpp Execution, Agent Tool Reliability, Quantization (GGUF/AWQ)
Core: Python 3.12, C++20, Rust, FastAPI, PyTorch
Infra: Docker, PostGIS, pgvector, Linux Kernel (x86_64)
Status: Available for high-impact contracts & remote engineering roles
Branch growth dynamically driven by verified code commits across repositories
| Engine / System | Architecture & Focus | Core Stack | Status |
|---|---|---|---|
| VulkanIlm | High-throughput local LLM inference wrapper utilizing Vulkan backends for consumer/non-CUDA GPUs via llama.cpp. |
Python C++ Vulkan GGUF |
 |
| Buildables AI Track | Production-grade LLM curriculum & execution code: Advanced RAG, structured tool-calling, and evaluation suites. | FastAPI pgvector LangGraph |
 |
| Urban-Intel Engine | Spatial intelligence platform integrating Vision Transformers, automated raster ingestion, and vector geometry. | FastAPI PostGIS PyTorch |
 |
|
|
|
System initialized. Deterministic pipelines, local weights, zero hallucinated state.