ML & Agentic AI Engineer — Chandler, AZ

Hello, I'm
Hector.

Agentic AI Engineer / Data Pipeline Architect / Backend Engineer

Chandler, AZ

10 Years at Intel

MS Computer Science

Discover My Work Download Resume

hector@hh-dev — zsh

$ whoami

ml/agentic ai engineer · 10 yrs @ intel

$ ./agent --status

[OK] etl latency 30s → 3s

[OK] agents voicevalet · mcp ×5

[OK] ms cs @ asu gpa 3.88

$ open projects/

About Me

I'm an engineer with 10 years of production experience at Intel, now building LLM-based tools, multi-agent systems, and scalable data platforms. I transitioned from process engineering in semiconductor manufacturing to software development, giving me a unique perspective on solving complex operational challenges with modern AI and backend technologies.

I design reliable agentic systems that combine models, tools, and data into end-to-end multi-step workflows — with explicit planning, structured outputs, tool use, failure handling, and monitoring and observability. From an AI phone agent that completes real calls to a collection of Claude MCP servers, I focus on shipping practical AI systems with sound latency, cost, and accuracy tradeoffs.

I'm completing my Master's in Computer Science at Arizona State University (GPA 3.88), with coursework in Agentic AI, Statistical Machine Learning, and Data Processing at Scale. I spent summer 2026 as a Data Engineering Intern at Honeywell, building LangGraph-based AI agents for refinery process monitoring on the Forge team. My decade at Intel taught me the importance of reliability and maintainable architecture — skills I now apply to building intelligent, scalable systems.

Download Full Resume

Technical Skills

Agentic AI & LLMs

Claude API OpenAI API MCP LangChain LangGraph LangSmith Ollama vLLM ReAct Function Calling Multi-Agent Orchestration Tool-Use Evaluation

Backend & Systems

Python FastAPI Docker Kubernetes PostgreSQL SQL Server Airflow Databricks CI/CD C# .NET

ML & Evaluation

scikit-learn Pandas NumPy τ-bench WildToolBench pass^k Metrics

Dev & Infra

Git GitHub Claude Code CLI SLURM HPC (A100 / AMD ROCm) VS Code

Supporting

JavaScript TypeScript React Angular D3.js

Education

Master of Science in Computer Science

Arizona State University | Expected December 2026

GPA: 3.88/4.0
Relevant Coursework: Agentic AI, Statistical Machine Learning, Artificial Intelligence, Data Processing at Scale, Advanced Operating Systems, Data Structures and Algorithms, Applied Cryptography, Data Visualization
CSE 598 Research (τ-bench): Evaluated agentic tool use across Qwen3 and Gemma4 model families using τ-bench tasks, error taxonomies, and pass^k metrics on vLLM, SLURM, and HPC clusters
Focus: Agentic AI, distributed systems, and backend architecture

Bachelor of Science in Chemical Engineering

Arizona State University | May 2015

GPA: 3.65/4.0
Foundation in analytical problem-solving, process optimization, and data analysis
Transitioned technical skills from manufacturing to software engineering

Professional Experience

Data Engineering Intern

Honeywell

May 2026 – August 2026
Atlanta, GA

Developed LangGraph-based AI agents on Honeywell's Forge team that automate trend detection and root-cause analysis over refinery process data — traced and evaluated in LangSmith — reducing manual investigation time for operations teams
Built an agent-assisted workflow supporting process engineers in sustaining engineering, translating plant-floor reliability requirements from a decade of semiconductor process experience into agent design decisions
Wrote Python and SQL analyses over large operational datasets in a Databricks environment spanning Azure and GCP data lakes

Software Application Engineer

Intel Corporation

June 2020 – July 2025
Chandler, AZ

Designed end-to-end architecture for Intel's global EDI manufacturing data platform, integrating Python prototypes, Angular/D3/Highcharts visualizations, and backend services for thousands of engineering users
Re-architected a legacy Python/Oracle ETL stack into C#/SQL Server, making cost/performance tradeoffs that improved scalability, maintainability, and long-term operational efficiency
Optimized database structures and query paths across services, reducing data transfer latency from 30+ seconds to under 3 seconds for production engineering workflows
Built Airflow, Docker, Python, and SQL Server pipelines with clear failure handling and reduced manual intervention, improving reliability of recurring manufacturing data workloads
Maintained globally distributed production infrastructure, resolving bugs and enhancements while strengthening monitoring, observability, and availability for manufacturing teams
Led code reviews and Python standards for the EDI development team, using LLM-assisted coding and structured review habits to improve quality, testability, and defect prevention

Photolithography Process Engineer

Intel Corporation

June 2015 – May 2020
Chandler, AZ

Sustained high-volume semiconductor manufacturing across Nikon and ASML scanners, balancing throughput, yield, cycle time, and cost through data-driven process control
Reduced defectivity mechanisms and scrap through targeted engineering solutions and tighter process control
Developed and deployed automation tools — 1-click reports, SQL queries, JSL and Python scripts — to extract, process, and visualize large datasets from legacy and modern systems
Trained and mentored new engineers on semiconductor equipment and photolithography processes, accelerating team ramp-up

Featured Projects

Showcasing agentic AI systems, production-grade backends, and data engineering solutions

Code

LLM Training From Scratch

Training language models from first principles — backprop to GPT-2 pretraining to QLoRA fine-tuning, following Karpathy's Zero to Hero. A lab notebook of code-alongs and benchmarked training runs on a local RTX 5090 and rented cloud GPUs.

PyTorch CUDA nanoGPT QLoRA Jupyter

Code

CodeLight

A minimal, native code editor for macOS built with Monaco Editor. Features syntax highlighting for 50+ languages, Git integration, multi-window support, and distraction-free editing.

Swift macOS Monaco Git Electron

Code

Voza

System-wide voice-to-text tool for macOS. Records speech via global hotkey, transcribes with OpenAI Whisper, cleans up with ChatGPT, and pastes polished text into any active app. Supports English and Spanish.

Python Whisper ChatGPT macOS

Code

VoiceValet

AI phone agent that makes outbound calls on your behalf to book appointments, get insurance quotes, and handle phone tasks. Built on Vapi, Claude, and Deepgram — sends a full transcript summary to your email via Resend when the call ends.

Python Vapi Claude Deepgram Resend

Code

Wedding Website

Elegant bilingual wedding website with rose gold theme. Features RSVP system, "Our Story" timeline, event details, and responsive design with beautiful animations.

React FastAPI PostgreSQL Tailwind Docker

Code

Money Review Page

Personal budget dashboard with interactive D3.js charts for spending trends. Features monthly/quarterly/YTD views, category breakdowns, and multi-user support. Deployed on Ubuntu with FastAPI backend, React frontend, PM2, and Nginx.

FastAPI React D3.js PostgreSQL Nginx

Code

Claude MCP Server Collection

A curated collection of 5 MCP servers for Claude AI: stock quotes, weather forecasts, live news headlines, Gmail management, and a PostgreSQL database connector. Extends Claude with real-time API access and email automation.

Python MCP Claude AI Gmail PostgreSQL

Code

LeetCode Practice

Daily coding practice repository featuring algorithm solutions with documented time/space complexity analysis. Organized by difficulty with test cases and optimization notes.

Python Algorithms Data Structures

Code

ConvergeAI

Multi-model AI consensus solver running GPT-5.2 and Claude in parallel, iteratively refining answers until they agree. Supports PDF, PPT, and HTML inputs with real-time cost tracking, disagreement analysis, and optional response caching.

Python OpenAI Anthropic PDF/PPT/HTML Cost Tracking

Code

Nexus Research

A production-grade multi-agent research pipeline that automatically decomposes a research question, retrieves information from public APIs, summarizes findings with confidence ratings, and produces a structured Markdown report — all orchestrated in a self-correcting loop.

Python AI Agents OpenAI Ollama LLM

Demo Code

Portfolio Website

The very site you're viewing right now! A personal portfolio showcasing professional experience, skills, and projects. Live at hectorhernandez.dev with modern design and smooth animations.

HTML CSS JavaScript

Hello, I'm
Hector.

About Me