josh laubach | Founder and Machine Learning Engineer

Josh Laubach Machine Learning Engineer | Reinforcement Learning, Computer Vision, LLMs | MS Data Science, UC Berkeley Math San Francisco Bay Area (US) 500 connections • 2,791 followers About I build AI systems that ship. My background is in applied mathematics (UC Berkeley) and data science (Boston University), with additional training in AI/ML from Caltech, and I spend most of my time at the intersection of machine learning research and product engineering. My recent work spans end-to-end autonomous vehicle pipelines using PPO and behavior cloning in CARLA simulation, computer vision robustness evaluation on nuScenes perception data, and LLM-powered products across healthcare, edtech, and consumer AI. Across six early-stage AI ventures, I have owned the full loop from model training to API deployment. The project I am most focused on right now is an AI math tutoring platform I am building from scratch. It uses LLM-driven hint generation with SymPy-verified solution validation, an adaptive difficulty engine tied to per-student KPI tracking, and a FastAPI plus React/TypeScript architecture designed for full autonomy. Skills I use regularly: Python, PyTorch, TensorFlow, Scikit-Learn, Hugging Face Transformers, fine-tuning LLMs, RAG pipelines, vector databases, reinforcement learning, causal inference, object detection, FastAPI, SQL, and SymPy. Open to Machine Learning Engineer and Applied Scientist roles in San Francisco. GitHub: github.com/joshlaubach Total Experience: 6 years and 6 months Experience Founder and Machine Learning Engineer - AI Math Tutoring Platform (Current) Jan 2025 - Present (1 year and 6 months) in San Francisco, California, United States Building a fully autonomous AI math tutoring platform from the ground up. - Architected LLM-driven hint generation with SymPy-backed solution verification, ensuring mathematical correctness at every step rather than relying on model plausibility alone - Built an adaptive difficulty engine th