Speaker media kit
William (MJ) Hill
Staff Software Engineer & Forward Deployed Engineer · AI · Atlanta, GA
william.mj.hill@gmail.com ·
linkedin.com/in/whill3 ·
github.com/William-Hill
Bios
Copy-paste ready. Use whichever length fits your program.
MJ Hill is a Staff software engineer and forward deployed engineer who builds production AI and data systems for regulated, customer-facing environments.
MJ Hill is a Staff software engineer and forward deployed engineer in Atlanta. Over 13 years, he has built data platforms, distributed systems, and applied AI software, including seven years under federal contract at a national laboratory and on Air Force and Space Force programs. He has spoken at AWS re:Invent, LeadDev, Strata, and the ACM Tapia Conference.
MJ Hill is a Staff software engineer and forward deployed engineer with 13 years of experience at Lawrence Livermore National Laboratory, New Relic, Meroxa, Zocdoc, and early-stage companies. He was the primary engineer on a Space Force Unified Data Library connector delivered under an Air Force Research Laboratory award. His recent work includes multi-agent systems, LLM evaluation, MCP and A2A integrations, and local model inference for regulated data. At RenderATL, he worked across sales, marketing, and engineering, closing $224,500 across 11 accounts while serving as the sole engineer on the platform those customers bought. He has delivered 36 talks, workshops, and panels since 2017 and developed the full curriculum for a workshop series that filled to a waitlist.
Talk topics I'm currently offering
Deploying AI where the internet isn't
What changes when regulated data rules out a hosted API. Drawn from building a local distillation and inference pipeline that kept FERPA-protected student data on institution infrastructure. Talk or 90-minute workshop.
Evaluation is the product
A practical evaluation stack for LLM systems: offline sets, online sampling, gold and honeypot tasks, inter-annotator agreement, model-graded review, and regression checks when models change. Talk or workshop.
Models propose, contracts dispose: why multi-agent systems need ontologies
A working multi-agent system returned confident, wrong answers even though each agent worked on its own. This talk traces the failure to the coordination layer and shows how an ontology gave the agents a shared vocabulary, constraints, provenance, inference rules, and refusal behavior. The same pattern applies to clinical, access-control, and manufacturing systems. Delivered at RenderATL 2026 (Breakout Stage 1, AI track). Talk, 25–30 min, with a live deployed demo. Also available as a workshop.
Beyond APIs: interoperable AI systems with MCP and A2A
What MCP and A2A solve, what they leave to the application, and how tool discovery, versioning, and scope affect a real agent deployment. Delivered at the Atlanta Developers' Conference, BlackTIDES bidWeek, and /dev/color. Talk or hands-on workshop.
From vibe coding to AI engineering
How to give a coding agent a useful spec, build reusable workflows around it, and enforce the tests and review steps it would otherwise skip. Based on a three-part series that filled to a waitlist. Workshop series, 3 × 90 minutes, or a single condensed session.
Constraint-driven design: a streaming architecture story
How hard memory, disk, and execution limits shaped an open-source Lambda log shipper. The session walks through the failed first design, the streaming replacement, and the tests that proved it worked. Talk. Originally presented at AWS re:Invent.
The Will Test
A nine-factor framework that helps underrepresented engineers judge whether a company can support them. Candidates, employees, and leaders can all use it as a concrete counterpart to the Joel Test. Delivered across five conferences. Talk, panel, or guided discussion.
Selected venues
| Year | Venue | Session |
|---|---|---|
| 2026 | RenderATL — Breakout Stage 1, AI track | The Put That Sh*t On Agent: Building Multi-Agent AI Systems That Actually Talk to Each Other — live deployed demo |
| 2026 | /dev/color MLPD Series | Vibe Coding with Guardrails: Rebuilding MajorKey with Claude Code (recorded) |
| 2026 | Atlanta Cloud + AI Conference | AI agents for open-source maintenance |
| 2025 | Atlanta Developers' Conference | Beyond APIs: Interoperable AI Systems with MCP and A2A |
| 2025 | GoWest (Forge Utah Foundation) | Real-time streaming data with Conduit |
| 2025 | BlackTIDES Data · bidWeek | MCP and A2A workshop (recorded) |
| 2024–25 | BITCON, Atlanta | AI workshop: LangChain, RAG, and agents (two consecutive years) |
| 2024 | The Bean Path | Using AI Agent Workflows to Create Your Personal Research Team — multi-agent orchestration and RAG (recorded, 71 min) |
| 2025 | Stats On Stats (podcast) | AI, Software Engineering & the Future of Tech — with a live RAG-over-NBA-statistics demo (recorded, 60 min) |
| — | Africans in Tech (podcast) | Episode 60 — Navigating Office Politics, Tech Layoffs & the AI Revolution (recorded) |
| 2024 | Data 4 Black Lives III | AI agent research workflows |
| 2022 / 2021 | MagnoliaJS · RenderATL | Global State Management in React (recorded) |
| 2020 | AWS re:Invent | How the Sausage is Made (recorded) |
| 2020 | LeadDev | Minimizing risks in code deployment, panel (recorded) |
| 2017 / 2019 / 2020 | ACM Richard Tapia Conference | Four accepted sessions, incl. a peer-reviewed technical panel (10 of 24 accepted) |
| 2019 | Strata Data Conference, O'Reilly / Cloudera | Evaluating company cultures (recorded) |
| 2017 | /dev/color in Motion | Moderator, Jeff Dean breakout session |
What attendees say
"Three Tuesday nights. Three workshops. One quiet shift in how I build… Part one was spec-driven development. The reminder that most AI coding problems are not coding problems, they are planning problems… I left after each workshop and immediately applied what I learned to refine the platforms I'd been working on."
Oneisha Freeman Rojales · AI & Social Impact Leader · RenderATL University attendee
Teaching credentials
- Curriculum developer and instructor, RenderATL University — 34-document curriculum, 164 registrations, first workshop waitlisted
- University of the Virgin Islands (2017, 2018, 2019) — invited summer instructor three consecutive years, both campuses; invited to join the faculty as an online adjunct
- Hackbright Academy — Senior Adjunct Instructor, 2015–2017
- Hidden Genius Project — session leader and curriculum designer · Black Girls Code and Kapor Center — volunteer instructor
- Ten public curriculum repositories spanning 2015–2025
On camera & remote delivery
I have worked on both sides of a live broadcast: on camera and behind the platform.
- Created and hosted a four-episode livestream interview series with conference speakers, totaling 2 hours 19 minutes of published content.
- Built the platform it ran on. The gated product at watch.renderatl.com handled ticket entitlements, authentication, playback, and the viewing experience for a 10,000-attendee conference.
- Broadcast television: live interview on 11Alive (NBC Atlanta) about AI and the future of gaming.
- Virtual conference delivery: presented the AWS re:Invent session twice, with a live product demo and audience Q&A.
- Nine published recordings in total, roughly five and a quarter hours.
Podcasts & interviews
| Show | Episode | Format |
|---|---|---|
| Stats On Stats | AI, Software Engineering & the Future of Tech — with a live retrieval-augmented-generation demo built over NBA statistics | Podcast · 60 min · recorded |
| Africans in Tech | Episode 60 — Navigating Office Politics, Tech Layoffs & the AI Revolution | Podcast · recorded |
| 11Alive (NBC Atlanta) | AI transforms the future of gaming | Live broadcast interview |
| RenderATL livestream series | Four-episode interview series with conference speakers — as host | Livestream · 2 hr 19 min |
Available for long-form interviews, unscripted conversations, and live technical demos.
I can also handle the production side of webinars, remote workshops, and recorded courses.
Booking & logistics
- Formats: conference talk (25–45 min) · workshop (90 min–half day) · multi-part series · panel · moderated fireside · podcast · livestream / webinar · recorded course
- Travel: based in Atlanta; available to travel domestically. International considered.
- Recording: happy to be recorded and to have material published.
- Lead time: four weeks preferred for a new talk; existing talks can run on shorter notice.
- Assets on request: high-resolution headshot, slide previews, and a session description tailored to your audience.
Booking: william.mj.hill@gmail.com