Ryan J. Kim

CS at Columbia. Systems, low-level programming, difficult problems.

Experience

Ramp — SWE InternFall 2026
Twitch — SWE InternSummer 2026
Roam — SWE InternSummer 2025

AI products adopted by 500+ users weekly. Go, TypeScript, React, MySQL, Redis, Kubernetes.

Columbia DitecT Lab — ML Researcher2024 - 2025

YOLOv8 recall +11% for NYC DOT. Physics-informed neural nets for traffic modeling.

Projects

Burn-TTT

LLM compiler for FPGA deployment via hls4ml. Test-time LoRA finetuning on synthesis feedback, beating random search and RF baselines.

rl4cuda

GRPO RL on Qwen2.5-Coder-14B for optimized CUDA kernels. Staged reward pipeline on A100 cluster.

Multithreaded Web Crawler

Concurrent C++20 crawler with inverted index, thread-safe queues, per-domain rate limiting, SHA-256 dedup.

Quantum ML for Dementia Detection

QCNN outperforming classical methods on MRI/CT scans. Published in IEEE.

xBeat

Voice-first DJ system with Grok AI. Natural language mixing and transitions.

Skills

C/C++, Python, Java, Go, TypeScript, SQL. React, PyTorch. Docker, Kubernetes, Redis, AWS.

Education

Columbia University · B.S. Computer Science

4.0 GPA · Egleston Scholar · Expected 2028