What I build

• I build web applications, data pipelines that move information, and integrations that connect systems.
• My work includes AI tools, testing, making software available to users, and support.

Who I work with

• I work with scientists, developers, healthcare teams, educators, and conference organizers.
• Their knowledge helps me decide what to build.

What I do now

• At RenderATL, I work on engineering and technical revenue.
• Through /dev/color’s Gates engagements, I work with institutions to define problems and deliver software.

Nevados · Solar energy

Built tools to process site data and generate solar-panel rotation schedules, plus visualization features for the customer web app.

Squared2020 · Sports analytics

Built Dr. Justin Jacobs’ NBA Running Net Points application: processing play-by-play data and creating an interactive view of player performance during a game.

Honey2Cocoa · Beauty technology

Advised an AI beauty startup matching makeup to skin tone for people of color. Guided technical strategy and reviewed the development team’s computer-vision model and product-matching approach.

Mela AI · Personal AI assistants

Advised the CEO and COO on architecture and AI strategy. Built a proof of concept with collaborating AI agents covering areas such as finances, health, and relationships.

“Forward deployed”

• Work closely with the people doing the work, on-site or remotely.
• Their systems and operating constraints shape the assignment.

The engineering responsibility

• Build, integrate, test, deploy, debug, and document software.
• Use customer conversations to guide decisions throughout delivery.

The result

• Check that the software works in context and helps its users.
• Use deployment lessons to improve the product or the next implementation.

Software consultant · Hired by the university

• Investigate the problem and help the university choose an approach: improve its help pages, buy a tool, or build a system.
• If implementation is in scope, build the solution, connect university data, and test it with staff.
• Deliver the agreed result for the university; the project may span several products.

Vendor FDE · Sent by the software company

• Help the university make your company’s AI platform useful for enrollment support.
• Build the integration, test answers with staff, and help launch it in their workflow.
• Bring missing features and recurring problems back to your product team so the platform improves.

The skill I carried forward

With limited time, I had to think about which answer would change my next action. A long list of questions is less useful than a question that helps narrow the problem.

Questions you can practice

Who is affected? What can they no longer do? What changed? Which observation would help distinguish the likely causes? How will we know the service has recovered?

Nevados · solar engineering

I built tools to parse geospatial point files and generate solar-panel rotation schedules. Understanding what the data represented helped me judge whether the output was useful.

Squared2020 · basketball analytics

Dr. Justin Jacobs supplied the Running Net Points methodology. I built data processing, an API, and an interactive application to explore it. An API lets software request information from another system.

Honey 2 Cocoa · learning an unfamiliar domain

I came in with no knowledge of makeup. I asked questions to understand how the company matched products to skin tones and what made a match useful. That understanding helped me advise on technical strategy and review the development team’s approach.

What carried forward

• From New Relic: ask which answer will change the next action.
• From consulting: learn the domain, agree on the problem, and earn trust.

Gates engagements · External institutions

• Through /dev/color, I work with university and foundation representatives on workflows, requirements, and prototypes.
• I deliver software, diagrams, and documentation. The Purdue example next shows that process in practice.

RenderATL · Inside the organization

• I work with conference teams on registration, streaming, and operational needs.
• I also use technical education and speaking to support the conference’s relationships and growth.

The concrete activities

I developed RenderATL University’s three-part workshop series, hosted four speaker interviews, and included a RenderATL invitation in my WeAreDevelopers presentation.

The purpose

These activities demonstrate technical education, build relationships, and create opportunities for people to consider participating in RenderATL.

How I would measure the contribution

I have not measured how many ticket sales came from external talks. Referral links, registration sources, or qualified conversations could help evaluate that contribution.

Skills to build

Practice functions, data structures, and error handling. Read and write API requests. Use Git, write tests, and trace a bug from its symptom to its cause.

Where I used them

At New Relic, I used Python to process and send log data. At Meroxa, I built connectors in Go. The languages changed; reading code, debugging, and checking behavior carried across.

A project to practice

Build a small registration API that accepts a name and email. Validate input, reject duplicate registrations, and write tests for successful requests and invalid data.

Skills to build

Use SQL to query and join tables. Model records and relationships. Read JSON or CSV, call an API, and check for missing or duplicate data.

Where I used them

At Meroxa, I built connectors that move data between systems. For Squared2020, I processed basketball play-by-play data and stored it for an interactive application.

A project to practice

Load a small public dataset into a database. Answer a question with SQL, expose the result through an API, and test missing fields and duplicate records.

Skills to build

Draw the path from user request to stored data. Decide where rules live, who can access each operation, and what happens when a dependency fails.

Where I used them

At New Relic, we processed S3 log files in pieces to fit the service’s resource limits. At RenderATL, registration needed capacity rules and streaming needed ticket-access checks.

A project to practice

Draw a registration form, its API, and its database. Show where the seat limit is enforced. Explain how duplicate requests, simultaneous bookings, and failures are handled.

Skills to build

Call a model through an API, give it relevant context, and connect only the tools it needs. Define what it may read or change and when a person must decide.

Evaluate the behavior

Create sample inputs with expected results. Check accuracy, unsupported answers, and failure behavior. Compare changes against the same examples instead of trusting a convincing demo.

A project to practice

Build an assistant over a small set of public documents. Write ten questions and expected answers, including questions the documents cannot answer. Check its answers and references.

The complication

People sometimes cancel. The organizer wants to avoid overbooking and know whom to contact when space opens.

Your task · 2-minute exercise

After the setup, take 70 seconds to write two questions that could change the design, a smallest useful version, and one test. Then we will hear a brief answer.

A constraint on your answer

Consider the organizer’s time and authority as well as the software. Use an existing tool, a process change, or custom code, and explain why the choice fits.

Questions and assumptions

• Can the organizer manage seat offers manually? How many registrations are expected?
• Assume a small event and an organizer who agrees to manage cancellations and offers.

First version

• A form adds entries to a sheet as waiting. Only the organizer changes status.
• Offer a seat only when confirmed plus reserved places total less than 30.
• Reserve an offer for 24 hours. Acceptance confirms it; a decline or expiry releases it.

Validation

• Test a full room, cancellation, acceptance, and expiry. Keep confirmed plus reserved places at or below 30.
• Match messages to the sheet. Release cancelled or expired places before making the next offer.

Read one primary source

• Read an official tutorial, release note, or documentation update for a tool you use.
• Identify the change, the problem it addresses, and its limits.

Run one small experiment

• Use sample data in a project you understand.
• For AI, compare output with an expected result and record a failure case.

Share what you learned

• Write a short note or show a classmate.
• Explain what worked, what you checked, and when you would use it. Investigate their questions.

S · Situation

Set the context. “The organizer could not tell who had a confirmed seat and who was waiting.”

T · Task

Name your responsibility. “I needed to give the organizer a reliable list before next week’s event.”

A · Action

Explain what you did and why. “I agreed on status rules with the organizer, updated the sheet, and tested cancellations and seat offers.”

R · Result

Describe what you observed. “In our tests, the list stayed within capacity and showed the correct status after a cancellation.”

Give people something to know

• Show a project, a small demo, or a problem you solved.
• Use a class discussion, meetup, community chat, or short post to make your skills visible.

Lead with value

• Listen to what someone is working on; share a useful example or answer a question you understand.
• Be clear about what you know and what you are still learning.

Keep the connection going

• Follow up with the resource you promised or explain how their advice helped.
• Show interest when you do not need a referral. Relationships take time and attention.

Keep building

Practice programming, testing, and debugging on a real workflow. Use AI where you can check its work.

Keep talking to people

Ask about their work, share something useful, and follow up. Let your projects and explanations give people a reason to remember your skills.

Stay in touch

William “MJ” Hill · mjhilldigital.com · github.com/William-Hill