I am interested in understanding and engineering the systems that make modern artificial intelligence fast, scalable, efficient, and reliable.
I am a Computer Science graduate and software engineer interested in the intersection of computer systems and artificial intelligence.
My long-term technical direction is AI Systems Engineering, particularly the infrastructure underneath large-scale training and inference. I am interested in problems involving GPU architectures, parallel programming, inference runtimes, distributed computing, memory systems, scheduling, communication, compilers, and performance optimization.
I like understanding computers across abstraction layers, from mathematics and algorithms to processors, operating systems, distributed runtimes, and large-scale AI infrastructure.
My approach is simple:
Build it. Measure it. Profile it. Break it. Understand why. Make it faster.
I am particularly interested in understanding systems such as vLLM, SGLang, PyTorch, CUDA, Triton, NCCL, distributed training frameworks, inference runtimes, and ML compiler stacks from the inside rather than treating them purely as APIs.
A retrieval-augmented LLM system for geospatial analysis integrating language models with geospatial tooling and external data sources.
The system explores:
- embedding and reranking services
- vector retrieval
- retrieval-augmented generation
- FastAPI-based services
- PyQGIS and WhiteboxTools integration
- satellite and geospatial data access
- Docker-based deployment
- cloud-hosted AI services
A processor microarchitecture implemented in VHDL capable of executing a subset of the ARM instruction set.
The implementation explores:
- ALU design
- register files
- program and data memory
- arithmetic and logical instructions
- branching
- load/store operations
- multiplication
- condition handling
- processor simulation and testing
This project reflects my interest in understanding computation below the software abstraction layer.
A hands-on operating systems project centered around implementing a Linux character device driver as a loadable kernel module.
The project explores:
- Linux kernel modules
- character devices
- kernel-space programming
- device-driver interfaces
- kernel architecture
- low-level operating system behavior
A chess engine implemented in Java for experimenting with classical search and decision-making algorithms.
Includes:
- alpha-beta pruning
- iterative deepening
- game-tree search
- board evaluation
- search optimization
Inference 路 Model Serving 路 Distributed Training 路 Distributed Inference 路 Scheduling 路 Memory Management 路 Performance Profiling
CUDA 路 GPU Architecture 路 Kernel Optimization 路 Parallel Algorithms 路 Memory Hierarchies 路 High-Performance Computing
Operating Systems 路 Distributed Systems 路 Computer Architecture 路 Networking 路 Compilers 路 Runtime Systems
Deep Learning 路 Reinforcement Learning 路 NLP 路 Computer Vision 路 Reasoning Models 路 World Models
I maintain a personal site where I document projects, technical explorations, notes, and experiments:
I am gradually building it into a technical knowledge archive covering computer science, systems, artificial intelligence, and the engineering underneath modern AI.