Sergio Zavala's Portfolio

About Me

Hey! I’m Sergio, a first-generation Latino engineer and a Computer Science graduate of CSU Monterey Bay. These days I work in agricultural robotics at Agrobotics, where I lead a field operations team and get to see robots doing real work outdoors every day.

Before that, I was an NIH NHGRI scholar at UC Santa Cruz building large-scale genomics software, and I built ML systems for summarization and education. I’m still finding my research direction, but I keep coming back to perception and robot learning for robots that work in the real world, from farm fields to planets and oceans. In January I start Georgia Tech’s OMSCS in perception and robotics.

Outside of tech, you’ll find me hiking, running, powerlifting, playing guitar, or behind a camera.

Experience

Data Operations Supervisor

July 2026 – Present

Agrobotics · Salinas, CA

Lead a field operations team
Develop and maintain operating procedures and quality checks
Work closely with the engineering team

Data Collection Engineer

April 2026 – July 2026

Agrobotics · Salinas, CA

Operated robotic systems in the field
Troubleshot hardware and software issues on-site alongside engineers

Student Assistant / Software Engineer

June 2025 – Present

Computing Talent Initiative · Remote

Support students asynchronously, helping them debug code and stay on track with their learning goals
Contribute to CTI-sys, CTI's backend data processing system, including a sanitization function that guards against malicious input

Software Engineer

February 2025 – June 2025

Computing Talent Initiative · Remote

Designed an AI grading pipeline that gives rubric-aligned feedback on ~100 students' blog posts about their CodeDay open-source experience
Cut feedback turnaround from weeks to about 10 minutes for a small staff

GitHub ↗

Open-Source Software Engineer Intern

October 2024 – November 2024

CodeDay × EBSCO · Remote

Built an AI summarizer for arXiv papers by fine-tuning a Longformer Encoder-Decoder (LED) model on 1,500 scholarly articles
Built a pandoc preprocessing pipeline over arXiv's S3 bulk data and a Flask API that summarizes a paper from its URL

Write-up ↗

Open-Source Software Engineer Intern

June 2024 – August 2024

CodeDay × Oppia · Remote

Fixed bugs on Oppia, an open-source learning platform with 1M+ users, in a team of three mentored by a UnitedHealth engineer
Restored music playback in lessons and prevented duplicate tags using Angular and JavaScript

Write-up ↗

Open-Source Software Engineer Intern

January 2024 – February 2024

CodeDay × Open Energy Dashboard · Remote

Contributed to OED, an open-source energy dashboard with 10k+ users, in a team of two mentored by a Hubtel engineer
Wrote a Mocha/Chai test that checks reading accuracy between a React component and the PostgreSQL database

Merged PR ↗Write-up ↗

Projects

A few things I've built, with the results to show for them.

ResearchBuddy

Team of 2 · lead contributor

2.8× ROUGE-1 vs. the pretrained model

AI summarizer for arXiv papers. Fine-tuned a Longformer (LED) transformer on scholarly articles, built a pipeline that pulls PDFs and LaTeX from arXiv's S3 bulk storage, and served summaries through a Flask API.

PyTorch
Hugging Face
NLP
Flask
AWS S3

GitHub ↗

CAPTCHA Solver

Team of 4 · built the best-performing model

52% per-character validation accuracy (baseline 19%)

Computer vision model that reads noisy 5-character CAPTCHAs. Fine-tuned VGG16 with data augmentation and five classification heads, one per character.

TensorFlow
Keras
Computer Vision
Transfer Learning

GitHub ↗

160+commits across the API and mobile app

GaelCraves

Team of 4 · lead contributor

160+ commits across the API and mobile app

Food-ordering platform for friends business. A Spring Boot API with JWT and OAuth login, Stripe payments and PostgreSQL, plus a React Native mobile app with ordering and admin tools.

Java
Spring Boot
PostgreSQL
React Native
Stripe

Backend ↗Mobile app ↗

NIH NHGRI Undergraduate Researcher

University of California, Santa Cruz

June 2025 – May 2026

Pangenomics
Algorithms
Bioinformatics

Built a Python tool that traverses pangenome graphs to map gene and transcript annotations onto every genome path, removing linear-reference bias
Used interval-tree indexing for fast region lookups, targeting the 462-assembly Human Pangenome Reference Consortium graph
Presented at NHGRI, SACNAS NDiSTEM and the CSUMB Summer Research Symposium

GitHub ↗

Undergraduate Researcher

California State University, Monterey Bay

February 2026 – May 2026

NLP
Model evaluation
RAG

Built an NLP pipeline that turns noisy course-evaluation comments into sentiment labels, themes and summaries
Benchmarked RoBERTa, DistilRoBERTa, Gemma and Llama across CSUMB and UCLA datasets, with metrics for insight quality and generalization

GitHub ↗