Avinh Huynh

CS grad from SFSU. Systems engineer at Arch Insurance. Starting an M.A. in Psychology (Mind, Brain, Behavior) at SFSU in Fall 2026.

Machine Learning

Cognitive science research at SFSU RADLab with Dr. Suri. Applying ML to study how we represent semantic memory.

DevOps

Systems engineer at Arch Insurance. Build automation and internal web tools for the team.

Full Stack

Build solutions to problems in my day-to-day, and craft customizable UX/UI that enhances offerings for clients.

Featured work

$100K/year

saved by automating cloud cost reporting and savings recommendations, replacing an external vendor with an internal tool.

AzurePower BIIBM CloudabilityPython

DevOpsWeb development

Automated FinOps Reporting

Systems Engineer at Arch Insurance

Internal cloud cost reporting and savings system, fully automated and customizable per business unit.

Customizable per business unit — each unit defines its own scope, metrics, and reporting cadence

Web GUI — non-technical stakeholders run reports on demand instead of filing tickets

Full Stack

Flow Immersive Website

Web Intern at Flow Immersive

Lead developer on Flow's public-facing Next.js website. Built the full site end-to-end — blog, admin dashboard, and supporting tooling — designing the UX/UI in close collaboration with the founder and other stakeholders.

Next.jsReactFlow Editor

Visit site

Automated Data Synchronisation

Cross-platform business continuity across hundreds of applications between two enterprise systems, on a schedule.

ArdoqFusionPythonREST APIs

DevOpsWeb development

Ardoq ↔ Fusion Automation

Systems Engineer at Arch Insurance

A scheduled Python script that merges records between two enterprise systems so architectural data stays in sync.

Detects broken records and dependencies — surfaces issues automatically so the ops team can triage them before they cascade

Web GUI — ops team approves new integrations manually and reviews flagged records

Machine Learning

ML Model Visualizer

Computational Cognitive Science Researcher at SFSU RADLab

A web tool for stepping through applied ML models — IAC, Rumelhart-style neural networks, and other semantic memory simulations. Built to help students build intuition for how these models actually work by watching each training step run in the browser.

Next.jsNeural NetworksCognitive Modeling

Visit site Source

Skills

Languages: PythonJavaScriptHTML/CSSSQLJavaC
Web: ReactNext.jsSpring BootNodeExpressThree.js
ML: PandasNumPyMatplotlibScikit-learnPyTorchTensorFlow
Cloud & Tools: AWSAzureGit/GitHubLinux (WSL2)DockerPostmanNginx
Databases: MySQLPostgreSQL

Experience

Dec 2023 - Present

Systems Engineer

· Arch Insurance — Infrastructure Analytics

DevOps

Eliminated $100K/year in external vendor spend by automating cloud cost reporting and savings recommendations with Python, Azure, Power BI, and IBM Cloudability.
Kept architectural data continuously synced across hundreds of applications by scheduling a Python service that merges records between two enterprise systems and surfaces broken dependencies for triage.
Unblocked non-technical stakeholders and the ops team by building web GUIs they can use to run reports on demand and approve integrations manually.

AzurePower BIPythonIBM CloudabilityArdoqFusionSnowflake

Feb 2026 - Present

Research Assistant

· SFSU RADLab — Dr. Suri

Machine Learning

Apply machine learning to study how humans represent semantic memory, working with Dr. Suri to implement and evaluate neural network models that mirror cognitive processes.
Help students build intuition for ML models by building web-based visualizers that show each training step — IAC, Rumelhart-style neural networks, and semantic memory simulations.

PythonPyTorchNext.jsNeural NetworksCognitive Modeling

Jun 2025 - Present

Web Intern

· Flow Immersive

Full Stack

Shipped Flow's public-facing website end-to-end as lead developer, covering the blog, admin dashboard, and CRUD content management on Next.js.
Designed the UX/UI in close collaboration with the founder and other stakeholders, then implemented every piece myself.
Turned business requirements into working features by translating direct conversations with stakeholders into shipped components and admin tooling.

Next.jsReactFlow Editor

Feb 2025 - May 2025

Research Assistant

· SFSU — Dr. Timothy Sun

Machine Learning

Produced verifiable solutions and pattern insights for Jungerman's Ladders under specific constraints by building Python scripts to solve and visualize valid configurations.

PythonPygameAlgorithm Design

Education

Expected 2028

M.A. Psychology — Mind, Brain, Behavior

· San Francisco State University

Concentration in Mind, Brain, and Behavior — cognitive neuroscience and computational models of cognition.

Spring 2026

B.S. Computer Science

· San Francisco State University

Minor in Computing Applications.
Certificate in Data Science and Machine Learning for Biotech.

Certifications

Jul 2025

AWS Solutions Architect Associate

· Amazon Web Services

View credential

Apr 2025

AWS Certified Cloud Practitioner

· Amazon Web Services

View credential

Personal projects

Hosting these onprem because I wanted to pretend I was an early-2000s startup; both are offline right now cuz I'm out of storage and saving up for a DAS lol.

Full Stack

SporkMoney

Jan 2026 - Feb 2026

I wanted to be less poor and know exactly how much I could afford to blow each day. So I built it. Plaid integration syncs all my accounts, and projection math handles daily allowance, savings runway, and emergency-fund timelines.

Next.jsPrismaPlaid APIDocker

Visit site GitHub

Full Stack

Streamour

Dec 2025 - Jan 2026

The streaming sites I was using were slow, kept buffering, and the subtitles were never in sync. So I built my own. MKV-to-HLS conversion for desktop and mobile, cookie auth, middleware-protected routes, and partial content delivery for clean seeking.

Next.jsPrismaHLSDockerNginx

Visit site GitHub

You made it to the bottom

Let's talk.

If anything on this page interested you — full-stack development, applied machine learning, or cloud architecture — I'd love to hear from you. The LinkedIn button is the fastest way to reach me.

Message me on LinkedIn Email GitHub

— Avinh Huynh · San Francisco