Projects
Technical builds, operating ideas, and communication projects—shown with their context, not just their output.
Personal platform · 2026
This website is entirely built by me—from its visual system and information architecture to the pages and responsive implementation. It brings my work, writing, and projects together in one place.
AWS Builder · Weekend Annoying Task Challenge · 2026
An AI writing assistant for drafting thoughtful LinkedIn comments and replies without handing over control of your account or voice.
Useful writing support that leaves the final word—and the send button—with you.
Comment Agent takes a LinkedIn post or comment you provide, then creates three distinct reply drafts for you to review, edit, and copy. It does not connect to LinkedIn, publish content, or automate engagement.
A user-controlled watchlist adds a small agentic workflow: items you choose to save can be drafted overnight and sent to you by email. If you save nothing, the agent stays quiet. The project is built with a static web interface, API Gateway, Lambda, Bedrock, DynamoDB, EventBridge, and SES.
Status: Deployed AWS prototype; source and a demo are available.
Read the AWS Builder Center article ↗Try Comment Agent ↗View the source on GitHub ↗AWS Builder Center profile ↗
AWS Productivity Challenge · 2026
A weekend sprint exploring a cloud-based system for turning unstructured thoughts into useful, organized action.
An experiment in turning a messy brain dump into a clear, useful next step.
BrainDump explores a simple idea: getting thoughts out of your head is only useful if you can make sense of them afterward. The project uses AI to help organize unstructured notes into actionable groups, giving the user a more manageable place to begin.
Built during an AWS productivity challenge, it focuses on making the first step feel lightweight: capture freely, then review and shape the suggested organization before acting on it.
Status: AWS productivity challenge prototype.
Hugging Face · 2025
The original scrollable portfolio and service site that combined local work, education, endurance, and early AI experiments.
The first digital home for a local-services practice built on trust and range.
Faber Solutions began as practical, person-to-person work across Northern California: property care, pet and house sitting, assembly, transportation, and whatever useful problem needed solving. The Space turned that broad body of work into a single public story that clients and collaborators could understand.
Built as a static Hugging Face Space with a DeepSite-assisted workflow, it combined services, personal background, education, endurance, and early AI experiments in one scrollable experience. It became the precursor to the more focused portfolio and writing system on zacfaber.com.
Status: Live original Space and expanded case study.
Explore the live Hugging Face Space ↗Read the Faber Solutions case study ↗
Interactive prototype · 2025
A guided-reflection prototype that turns someone’s memories, identity, and current goals into visual prompts for reframing a difficult moment. Built around the idea that a person’s own story can help them see a path forward.
A guided reflection concept for finding a steadier next step.
Solace explores how AI might help someone pause, put words to a difficult moment, and reconnect with a personal source of meaning. It uses a person’s own chosen memories, values, and goals as material for reflection—not to diagnose them or replace a mental-health professional.
The early prototype experiments with turning personal context into a visual, guided reflection. The next evolution is a more thoughtful AI wellness experience: clear about its limits, respectful of what people choose to share, and oriented toward one small next step.
Status: Early interactive prototype; the broader product direction is still being developed.
Personal data story · 2026
A magazine-style analysis of a personal Apple Music library, exploring artist rotation, listening concentration, and how taste changes over time. The report turns a full-library data analysis into an editorial story.
A more human way to explore the patterns in a music library.
The Rotation Review looks beyond a top-songs recap. It explores what keeps returning to someone’s listening, how their rotation shifts, and what patterns might be worth noticing. The product direction is to make those observations easy to ask questions about—with AI helping explain the data while showing the evidence behind each observation.
The current work includes a local browser-based analytics prototype and an editorial report. The next evolution is a privacy-conscious, sample-first experience: people can try it with fictional data, or explore their own library with clear control over what is processed and shared. It describes listening patterns, not unsupported claims about someone’s identity or mental health.
Status: Working analytics experiments and an editorial prototype; the public product experience is still being shaped.
AI systems practice · 2026
A collection of patterns for turning open-ended goals into bounded, reviewable workflows—using agents where they help and keeping human judgment in the loop.
I build agent workflows around useful outcomes—not autonomy for its own sake.
I start by defining the job, the trusted information it can use, and what a good result should look like. Then I shape the workflow into focused steps, giving each agent only the context and tools it needs. When tasks are independent, I may let agents investigate in parallel; a coordinating step compares their work and resolves gaps.
The workflow should make its reasoning and handoffs inspectable. Important actions need clear approval points, and the final result should be something a person can verify—such as a checked answer, a proposed change, or a tested artifact. I test the running system, not just the prompt or code, and tighten boundaries when real behavior exposes a failure mode.
Status: An evolving practice of designing, testing, and improving AI workflows around clear goals and human review.
Want to learn more? Ask me about how I design and build agentic workflows.