AI Developer Advocate • Open Source • Speaker • Builder
I help developers build with open models, through code, workshops, and talks. I focus on LLMs, inference, and open source. Currently at Nebius.
Hi, I’m Sujee 👋
What I Work On
Developer Advocacy
Code, workshops, and community for AI builders.
Open Source
Open-source tools, evals, and benchmarks.
Talks & Workshops
Sharing what I learn about AI, LLMs, and inference.
Selected Work
Practical LLM Evals
Model visualizers, developer utilities for testing inference, and more.
ZebraBench
Benchmark LLM inference endpoints right in your browser: speed, latency, accuracy, and cost.
Allycat
End-to-end open-source RAG pipeline for website content.
Selected Talks
Recent Events
Latest Writing
Experimenting with Agaram: A Tamil-First Language Model
I ran Agaram 2.5B locally and tested 20 phrases in both Sri Lankan and Tamil Nadu Tamil. It translated 17/20 Tamil Nadu phrases and 12/20 Sri Lankan phrases correctly.
31 Years With Linux: From Compiling Kernels to Running AI Models
Linux 0.02 was announced 35 years ago. I’ve used it for 31 of those years - from recompiling kernels to running AI models locally.
Fine-Tuning vs. Distillation vs. Quantization, Explained with LEGO
Fine-tuning, distillation, and quantization sound similar but solve different problems. This LEGO explainer shows how.
What People Say
Peter J. Theobald, Akamai
“…one of the best I’ve attended. Clear, concise and hands-on, it really helped me understand all the components to building a LLM pipeline.”
Mike Chrabaszcz, Engineering Leader
“…a well organized, timely, and insightful workshop at the AI Summit.”
John Varghese, AWS Hero
“Sujee has a knack for explaining complex topics using easy to understand analogies.”