# Sergio W. Peterson

> Auto-generated from markdown source files in /content and /blogs.
> Source markdown links below are the canonical LLM context for this site.

## Canonical Links
- Website: https://www.sergiopeterson.dev/
- LLM index: https://www.sergiopeterson.dev/llms.txt
- Resume PDF: https://www.sergiopeterson.dev/my_resume.pdf
- github: https://github.com/SergioPeterson
- linkedin: https://www.linkedin.com/in/sergio-w-peterson/
- twitter: https://x.com/sergiopeter4
- instagram: https://www.instagram.com/sergiopeter2020/

## Source Markdown Files
- Profile source: https://www.sergiopeterson.dev/llms-src/content/profile.md
- Projects source: https://www.sergiopeterson.dev/llms-src/content/projects.md
- Experience source: https://www.sergiopeterson.dev/llms-src/content/experience.md
- Resume source: https://www.sergiopeterson.dev/llms-src/content/resume.md
- Blog source (The-Shadow-Agent): https://www.sergiopeterson.dev/llms-src/blogs/The-Shadow-Agent.md
- Blog source (Efficient-Graph-Neural): https://www.sergiopeterson.dev/llms-src/blogs/Efficient-Graph-Neural.md

## Identity
- Name: Sergio W. Peterson
- Location: Berkeley, CA
- Title: Software Developer | Data Scientist | Founding Engineer
- Short bio: UC Berkeley grad - ML, robotics, and full-stack systems.

## Bio
UC Berkeley graduate with a B.A. in Data Science focused on robotics and software engineering.
I specialize in building full-stack systems and AI pipelines from scratch, with a focus on machine learning, robotics, and large-scale infrastructure around LLMs, computer vision, and autonomous systems.
Currently building Scholaria, a daily personalized research reader for arXiv and other sources.

## Tech Stack
- Python
- C++
- JavaScript
- TypeScript
- React
- Node.js
- PyTorch
- TensorFlow
- OpenCV
- YOLO
- PostgreSQL
- Redis
- MongoDB
- AWS
- Docker
- Kubernetes
- GitHub Actions
- ROS

## Hobbies
- Collegiate Archery Team
- Guitar
- Weightlifting

## Quote
"You can lose 100 10,000 times. All you need to do is win once." - Sergio W. Peterson

## Projects
- corgi-insurance-claims-classifier
  name: Corgi Insurance Claims Classifier (YC S24)
  summary: Insurance claims classifier using GPT inference with vector retrieval.
  tech: Python, GPT, Vector DB, OpenAI API
  featured: true
- movie-recommendation-system
  name: Movie Recommendation System
  summary: Hybrid recommender using collaborative and content-based filtering.
  tech: Python, Flask, React.js, PostgreSQL, Elasticsearch, SVD, TF-IDF, Docker, AWS
  featured: true
  link.github: https://github.com/SergioPeterson/Movie-Recommendation-System
- voidformer
  name: VoidFormer - Custom Transformer
  summary: Spanish to English transformer with faster training and strong BLEU.
  tech: PyTorch, Transformers, NLP, Attention, BLEU Evaluation
  featured: true
  link.github: https://github.com/SergioPeterson/VoidFormer
- scholaria
  name: Scholaria (on hold)
  summary: Research reader app with AI summarization (on hold).
  tech: React Native, Node.js, PostgreSQL, Redis, Render
  featured: false
  link.github: https://github.com/Scholaria
- rate-my-anime
  name: Rate My Anime
  summary: Anime rating and recommendation app with ML-based taste classification.
  tech: React.js, Node.js, PostgreSQL, MongoDB, MyAnimeList API
  featured: false
  link.github: https://github.com/SergioPeterson/RateMyAnimeList
- custom-character-network-analysis
  name: Custom Character Network Analysis
  summary: Character network scraping and large-scale graph analytics pipeline.
  tech: Graph Algorithms, Data Pipelines, Network Visualization, Cloud Computing
  featured: false
  link.github: https://github.com/SergioPeterson/anime_network
- restaurant-review-score-prediction
  name: Restaurant Review Score Prediction
  summary: Yelp score prediction with sentiment plus ML features.
  tech: Python, SQL, VADER, Word2Vec, NN/RF/LR
  featured: false
  link.github: https://github.com/SergioPeterson/Yelp_score_pred
- voice-recognition-automobile
  name: Voice Recognition Automobile
  summary: Voice controlled car with real-time command detection.
  tech: PCA, SVD, Linear Regression
  featured: false
  link.github: https://github.com/SergioPeterson/voice-recognition-car

## Experience
- minerva-intelligence-ycx25
  role: Founding Engineer, First Hire
  company: Minerva Intelligence (YCX25)
  period: Current
  summary: Building AI-powered intelligence platform as founding engineer.
  link.company: https://www.tryminerva.ai/
- lexius
  role: CV/ML Intern
  company: Lexius (YCW2026)
  period: 2024
  summary: Improved production video classification quality and metrics.
- ai-racing-tech
  role: Simulation & Perception Researcher
  company: AI Racing Tech (UC Berkeley)
  period: 2023-2024
  summary: Worked on autonomy stack simulation and CI for racing systems.
- berkeley-speech-group
  role: LLM Poker Researcher
  company: Berkeley Speech Group
  period: 2023
  summary: Trained and evaluated an LLM poker decision system.

## Blogs
- Published: 1
  - Efficient-Graph-Neural | 2024-01-15 | Efficient Graph Neural Networks for Large-Scale Molecular Property Prediction
    read_time: 10 min read
    category: Research
    excerpt: Graph neural networks (GNNs) are a powerful approach for molecular property prediction, but applying them to large-scale datasets is still computationally demanding. In this work, I test FastMolGNN, an efficient GNN architecture that uses hierarchical pooling and sparse attention to handle millions of molecules. I evaluate FastMolGNN on the Open Graph Benchmark (OGB) and observe state-of-the-art performance on molecular property prediction tasks, while reducing training time by 40% compared to standard GNN baselines. My ablation studies highlight the importance of hierarchical pooling for capturing long-range dependencies in molecular graphs.
    source: https://www.sergiopeterson.dev/llms-src/blogs/Efficient-Graph-Neural.md
- Not published: 1
  - The-Shadow-Agent | 2026-02-16 | The Shadow Agent: Multi-Agent Architectures for Self-Correcting Autonomous Workflows | status=draft
    source: https://www.sergiopeterson.dev/llms-src/blogs/The-Shadow-Agent.md

## Notes
- This file is generated by scripts/generate-llms-txt.js.
- Edit content in /content/*.md and /blogs/*.md, then run npm run generate-data.
- Raw markdown is published under /llms-src/ for exact source fidelity.