# Krisztian Kovacs ML engineer turned CTO. I worked on LLM-based apps; covering the whole stack from infra to evals to frontend. ## Preferred Stack - Infra: AWS, SST, Seed, Vercel / OpenNext, PlanetScale, Turso, Opentelemetry - ML / LLMs: PyTorch, Unsloth, Pydantic !!, Predibase, Handrolled agents / evals, - Frontend: React / NextJS, SolidJS, Tailwind ## Work Experience ### CTO - [Noki](https://noki.so) **Oct 2021 - Present** - [See Noki Walkthrough](https://youtu.be/DHH_6qdsyUY) - Built LLM agents to research & qualify companies for better sales & marketing targeting. - Built in-house evals & LoRA finetuned LLAMA 3 models to match GPT4 performance. ### Deep Learning Engineer - [Greyparrot](https://www.greyparrot.ai/) **Jan 2019 - Apr 2020** - Implemented a PyTorch fork of Yolo V2 (from C / CUDA). - Built hardware prototype for recycling waste monitoring (incl. hooking up cameras, lights, chip, etc.). - Optimized object detection/tracking models (& data pipelines) for on-device (NVIDIA Jetson) inference. ## Side Projects ### AI company list A curated list of [AI products](https://aicompanies.dev) (both infra & applications). Scraped & classified by AI agents. ## Education ### New York University Abu Dhabi **2010 - 2014** Bachelor's Degree, Economics and Mathematics, GPA 3.8 ## Contact - kris@kk1694.com - [https://krisztiankovacs.com](https://krisztiankovacs.com)