# Justin Foley **VP of Engineering** | Columbus, Ohio *16 years building systems, teams, and the outcomes they drive* *design / build / ship / scale* Engineering executive leading 40+ engineers across five product lines at Later, an AI-native creator marketing platform. I meticulously craft operational conditions that cultivate joy and productivity, and tie engineering to business results, from ground-up platform rebuilds to board-level AI strategy. Founder instincts, operator discipline. **16+** years building · **100+** engineers led · **25+** products shipped Email: [justin@modweb.io](mailto:justin@modweb.io) · LinkedIn: [in/jfols](https://www.linkedin.com/in/jfols) · Agency: [modweb.io](https://www.modweb.io) · Resume: [resume.md](/resume.md) · Resume PDF: [resume.pdf](/resume.pdf) --- ## Philosophy > Complexity begets complexity, simplicity begets elegance. **Belief:** density over headcount **Substrate:** trust, autonomy, joy **Loop:** ship / measure / learn Five great engineers beat twenty good ones. Every time, per dollar. Top talent is the cheapest thing you can buy, so I hire slow, pay up, and hold the bar high. If you optimize for cost per head, I am the wrong hire. The 10x engineer is not a myth. Trust, autonomy, real problems: give people those and the output follows. I build the substrate before the sprint board, the conditions that make great work ordinary instead of heroic. Relationships are the foundation; the rest is plumbing. I sequence by cost of delay, measure in net profit impact, and shrink the batch until feedback comes fast. Ship the smallest bet that teaches you something. Kill what the data kills, fund what it proves. Expect me to push your thinking past where it stops. --- ## Gigs ### [Later](https://later.com) VP of Engineering | 2025 to present · Remote Promoted to VP in three months, the fastest in company history, after inheriting an org mid-turnover and rebuilding the system that produces engineering output. Lead 40+ engineers across five product lines, three layers deep to the IC, with director-, principal-, and architect-level direct reports; own engineering's AI strategy at the board level. Later is an AI-native creator marketing platform: $2.4B+ annual GMV, 3,000+ brands, and 180,000+ creators, backed by Summit Partners. **AI & Architecture** - Shipped [Later EdgeAI](https://www.prnewswire.com/news-releases/introducing-later-edgeai-the-industrys-most-advanced-predictive-intelligence-solution-302596310.html): semantic creator matching across 16M+ profiles on a custom SurrealDB knowledge graph. Sub-ms queries; collapsed 10-minute workflows to seconds. The company's first major AI capability and lead marketing feature: recentered the brand on AI and became our dominant answer to the AI question every 2025 client was asking, lifting deal closures and deal size. - Built the engineering intelligence platform across 20 teams: real-time health, quality, productivity, CapEx, and AI effectiveness. Killed manual spreadsheet audits; showcased to Summit Ventures and earned external investor validation. - Stood up a Go MCP server and docs monorepo (manifests, C4, cross-repo synthesis) queryable from any IDE, closing a SOC 2 architecture-diagram gap in the process. - Driving the agentic dev-tooling thesis as the company's engineering AI strategy. **Measurement & Business Impact** - Built Linearbot, an opinionated layer over Linear, GetDX, and AI tooling metrics that systematized our way of improving team flow. Engineers got transparent feedback and WIP-constraint protection from over-subscribing; product and business got velocity-based progress and projections. It took emotion out of prioritization, exposed unfocused "yes to everything" patterns, and drove laser focus across the product org. - Built an R&D capitalization methodology Finance adopted, surfacing $3M+ in EBITDA through improved CapEx categorization; GetDX called the approach "lightyears ahead." - Quantified AI adoption impact across an 80-engineer org (GetDX, Q1 2026): ~25K engineering hours reclaimed annually, roughly 12 FTE of capacity and ~$2.3M in loaded cost. - PR throughput up 73% and defect ratio down 2.3 points in a single quarter (GetDX, Jan to Apr 2026). **Operating Model & Culture** - Shipped the Engineering Operating Framework: WIP limits, protected capacity, technical veto, fixed-time/variable-scope. Put tech debt on the board agenda. - Replaced status meetings with agent-friendly writing and POCs: "a POC is worth a thousand meetings" is now org shorthand. - Operated engineering through the Mavely post-acquisition integration into the unified Later platform, a growth-equity-backed M&A environment. **Hiring & Talent** - Redesigned hiring end to end: killed take-homes, AI-graded screens, live coding with AI. Hundreds of interviews and dozens of hires; three in my first 90 days while onboarding. - Coach leaders, not just ICs: a director toward VP readiness, a staff engineer to promotion, a TPM into staff-executive positioning. - Treat retention as a systems problem (DX data, Culture Amp, financial modeling at the exec table); retained 100% of core critical talent through the turnaround. **Partnerships** - Founder-level partnerships with SurrealDB, Encore.dev, and GetDX. ### Shopmonkey Engineering Manager · Product Manager | 2023 to 2025 · Remote - Built enterprise messaging infrastructure (NATS streaming, AWS SES, Telnyx): real-time notifications, email, and SMS at scale. - Shipped a multi-provider payment platform (Finix, Stripe) and a subscription engine for multi-company franchises; ran hundreds of merchant migrations. - Turned around a high-risk v1 to v2 migration of the core platform thousands of repair shops run on, coordinating across departments; led technical due diligence on BNPL integrations (Sunbit, Affirm). ### Filevine Lead Senior Software Engineer | 2021 to 2023 · Remote - Led [Filevine's](https://www.filevine.com/) largest feature team (billing, timekeeping, activity, tags) on the regulated legal-practice-management platform; recruited and hand-picked the team, including a top company engineer and a star designer. - Rebuilt the tagging system end to end: swapped the engine mid-flight and migrated every customer. - Built the contenteditable engine behind Filevine's notes (caret management, TreeWalkers); led the frontend in Svelte with the org's first unit tests, Storybook, and a custom state-management pattern. ### iVueit CTO | 2018 to 2021 · Columbus, Ohio - Led a ground-up platform rebuild on Go, gRPC, IaC, and Svelte; cut over with minutes of downtime. - Migrated the legacy app to AWS and built a legacy-DB API that unlocked customer automation and new revenue. Redesigned admin and end-user flows to cut manual work and fraud. - Partnered with founders on sales, fundraising, and strategy. ### ModWeb Founder · Principal Consultant | 2014 to present · Columbus, Ohio One partner who sells, scopes, designs, builds, and ships. Since 2014 I have owned the full SDLC end to end across dozens of engagements: B2B sales and contracts, architecture, delivery, and the strategy work with founders, CEOs, and CTOs that decides what gets built and why. Zero to one is the specialty, carrying an idea through validation, architecture, and launch. The work spans utilities, fitness, civic, nonprofit, and social: utility account automation for [Nationwide Energy Partners](https://www.nationwidenergypartners.com), a Lean Six Sigma platform for [MaxPT](https://maxpt.net/), a realtime workout UI for [Complete Fitness & Performance](https://www.cfpcolumbus.com/), a [Jack Welch](https://en.wikipedia.org/wiki/Jack_Welch) interactive video wall at [Sacred Heart University](https://www.sacredheart.edu/), the MergeIn social video platform, a grants CRM for Culture Works, and web work for the [City of Westerville](https://www.westerville.org/). ### Earlier 2010 to 2014 - Dropoly: CTO and first technical hire, 2011 to 2014. Built a predictive home-energy model on NOAA data; pivoted into K-12 energy education sold through utility partnerships. - Epic Systems: Developer, 2010. --- ## Technical depth **Architecture & AI:** Knowledge graphs (SurrealDB), Vector embeddings, Agentic / semantic search, MCP, Distributed systems, Event streaming (NATS), gRPC **Cloud & Infra:** AWS, IaC, Docker, CI/CD **Languages:** TypeScript/JavaScript, Go, Bash **Data:** Postgres, MongoDB, CockroachDB, MySQL **Frontend:** Svelte (since 2019), React, Tailwind, Flutter **Leadership:** Org design, Engineering operating models, R&D capitalization, Hiring systems, Technical sales, Public speaking --- ## Community ### [Westerville Web](https://www.meetup.com/westervilleweb/) Founder | 2017 to present · Westerville, Ohio Monthly tech, design, and business meetup. 1000+ members, 35+ events, 47+ speakers. ### WestCowork Founder | 2018 to 2021 · Westerville, Ohio Built a coworking space in Uptown Westerville in an old Masonic Temple. --- ## Speaking ### Emcee 60+ Events | 2017 to present Grab the mic, energize the room. Available. ### Agentic DX Conference | 2026 · Vancouver, BC One path from spec to production, whether a human or an agent drives it. The substrate agents run on is the binding constraint, and good for people, decisive for agents. ### Measuring Engineering Impact Summit Ventures | 2026 · Private Invited talk in the Summit portfolio expert series on measuring engineers with an opinionated layer over Linear, GetDX, and AI tooling metrics. Summit CTO Tim Kohn said it raised the bar for future talks. ### Generate All The Things NewRX | 2018 · Newark, Ohio Generative app scaffolding in React, Redux, Meteor. ### Groovin with Bodymovin Westerville Web | 2017 · Westerville, Ohio Pure SVG animation in React with Bodymovin. ### Finite State Gumball Machines St. Paul School Code Club | 2017 · Westerville, Ohio Finite state machines via interactive gumball-machine exercise. Young minds, blown. ### Meteor Magic Pillar Plugged In | 2016 · Columbus, Ohio Technical inception via slide deck. Live deployment as performance art. --- ## Competition ### [Omnibus](https://www.columbusunderground.com/omnibus-bw1) GiveBackHack · 1st Place | 2016 · Columbus, Ohio Realtime COTA bus tracking MVP application. ### LocalFlow Startup Weekend · 3rd Place | 2015 · Columbus, Ohio Local guides build activity sequences for unique experience flows through a city. --- ## Education ### M.S., Computer Science University of Dayton · 2012 Focus: graph algorithms and computational reflection. ### B.S., Computer Science Math minor | University of Dayton · 2010 --- ## Agentic DX > Easy to generate is not simple to operate. DX is agent leverage. **Substrate:** built for agents **Loop:** spec → production Cheap code is not a simple system. The hard parts stayed hard: specification, design, debugging, integration, coordination. The substrate reinforces them, and sets the ceiling. These properties have always mattered for humans. Agentic engineering makes them critical. As the work shifts to agents, the substrate they run on becomes the binding constraint: how fast they can try, how safely they can fail, how far they go before a human steps in. Good for people, decisive for agents. One path from spec to production, whether a human or an agent drives it. **The work loop** 1. Pick up the issue; spin up a self-contained environment preseeded with realistic data. 2. Iterate with tight feedback until the spec is satisfied. 3. Open the PR: CI tests and a full-stack preview deploy automatically with E2E validation. 4. A human reviews and validates against the preview where it applies. 5. Merge, deploy, done. ### Spec **Work starts from a spec** *Agents invent the contract while coding. Reviewers chase a moving target.* Implementation and tests both come from it. The spec is the diff. ### Feedback **Feedback in seconds** *Slow feedback makes each attempt expensive. The loop stalls before it finds the answer.* Type errors, lint, and service contracts before it runs. Tests, hot reload, and output diffs as it runs. Try, fail, try again at the speed of LLMs. ### Reproducibility **Same inputs produce the same outputs** *Flaky builds hide real failures and trust erodes.* Hermetic toolchains, deterministic dependencies. Flaky tests are noise. ### Boundaries **Boundaries are contracts the build checks** *Each new boundary the agent crosses without a typed contract becomes a failure surface the build won't catch.* Boundary calls type-check at build time. Implementations swap behind the contract. The caller doesn't change when they do. ### Preview **Changes run in a faithful copy of production before they merge** *Bugs a preview would have caught reach production.* Every PR gets a full-stack environment. Production data branches per PR, the way code does. Bugs reproduce against real state; migrations and schema changes test against real volume. ### Gates **Change confidence runs at agent speed** *Agents ship faster than humans can review. Without gates that defend a merge on their own, confidence collapses and production absorbs the cost.* A gate suite runs in parallel on every PR against the preview: E2E, latency, memory, dollars-per-request, output quality. Green means go. Anyone with context can review first. ### Containment **A bad ship fails small and recovers fast. Retries replay safely** *Speed multiplies bets. Containment caps the pain; recovery caps the time. Without both, the business writes a blank check on every bad ship.* Consumers are idempotent. Topics are at-least-once. Feature flags as a primitive. Canary deploys by default. Automatic rollback when regression signals fire. ### Signals **Agents are only as good as the signals they read** *Bad signals, bad actions.* An agent investigating a failure reads the trace, logs, and error class as input for the next attempt. ### Knowledge **The system answers questions about itself** *Agents start cold. Docs drift. The system the agent reads about isn't the system it's working on.* The system is the source of truth, and it describes itself on demand. Decisions, incidents, and operational data join it on the same live surface. Past decisions and failures become context for the next change. ### Accountability **Actions attributed, costs measured, risks gated** *Without accountability, agent actions become anonymous. The business inherits the cost without inheriting the cause.* Agents read freely, write only in their own branch, and need a human on the risky moves. Every action is logged and replayable. Every AI request carries the feature, team, and customer it serves. *A beginning, not an end. A foundation to build on and a spark for the conversations that carry us into an agentic future.* ---