Brandon

Founding CTO and lead engineer. Built Pier's production AI platform from market thesis to live external usage in 12 months (launched February 4, 2026) so employers can see what makes each candidate qualified beyond AI-polished application materials; led a cross-functional team across engineering, design, marketing, and legal, and helped raise $1M+.

• Architected and built Pier's production real-time voice AI platform from scratch, shipping Hiring Lab and SmartApply: live candidate interviews, resume intake, candidate evaluation, explainable fit reports, employer dashboards, recruiting workflows (question, competency, and rating-scale generation, fit analysis, streaming assistant chat), and the infrastructure behind them, owning every layer from the Next.js/React frontend to the Python AI services and Postgres data model.
• Designed the evaluation architecture that turns live voice conversations and resumes into explainable assessments of role fit, differentiation, team fit, and interview readiness: an LLM-as-judge layer with employer-reconfigurable rubrics per question that cites transcript evidence, named criteria, and confidence scores behind every recommendation, so decisions stay traceable, defensible, and correctable by humans, keeping pace with bias-audit regulation such as NYC's Local Law 144.
• Built live interview sessions on LiveKit AgentsJS, Deepgram speech-to-text, ElevenLabs/OpenAI text-to-speech, and a multi-provider OpenAI, Claude, and Gemini LLM layer, with dynamic prompting, streaming, agent routing, transcript persistence, and latency tuning.
• Orchestrated deterministic, tool-calling agent workflows in Python with LangGraph, routing reasoning tasks across frontier and smaller models by complexity and handling context management (window size, caching) to hold latency and token cost in check.
• Built evals, with LLM-as-judge graders and spec-derived regression suites, to debug and tune prompt, model, and orchestration behavior across the pipeline, giving every change a repeatable quality check.
• Cut resume-extraction cost 91% by replacing image-based PDF processing with text extraction while improving reliability on incomplete and imperfect resumes; redesigned upload as an async pipeline (durable acceptance, background parsing) that replaced a roughly 60-second blocking flow with a sub-30-second candidate-facing target.
• Modeled voice-infrastructure economics and identified a 66% per-interview cost-reduction path, lifting projected gross margin from 58.5% to 85.9%.
• Roughly doubled onboarding retention (about 40% to 80%) and cut churn (about 60% to 20%) through PostHog product analytics and A/B experimentation driven by real usage signals, carrying the A/B-through-Z testing habit formed at OkCupid into Pier.
• Applied matching-systems experience from OkCupid, which facilitated 25M+ matches, to Pier's agentic candidate-role matching and compatibility-scoring architecture.
• Led technical strategy for a flexible evaluation platform extending beyond hiring into other requirement-governed decision workflows, including vendor qualification, admissions, and investor screening.
• Architected an AI-native development and release system combining multi-agent execution, autonomous end-to-end QA, evidence gates, and human-approval guardrails, cutting a four-to-five-day QA cycle to hours; spec-derived evaluation suites caught cross-role data-integrity failures before release.
• Instituted production-equivalent staging, a P1 stop-the-line policy, smoke tests, observability, and architectural review as standing engineering requirements, with persisted transcripts and PostHog instrumentation giving every interview and recommendation an inspectable trail.
• Owned company and product strategy from customer research through positioning, roadmap, signed customer commitments, and live deployments, using the evidence-first Strategic Proof method later open-sourced as Brandon's Strategy Notebook; authored investor- and diligence-ready technical and security narratives covering AI architecture, model-risk boundaries, human oversight, and data flow.
• Built production infrastructure-as-code (OpenTofu, Cloudflare, Vercel, GitHub Actions, Infisical, Cloudflare R2) across domains, environments, secrets, storage, and CI/CD, and enforced Clean Architecture boundaries across the TypeScript/React/Node.js product codebase and Python AI services (tRPC, Postgres) with Vitest, Playwright, Storybook, and semantic-release so agentic coding could scale safely, applying the public Clean Architecture guide for TypeScript/Next.js projects authored in 2025.
• Built Pier's web app on Next.js (App Router) with React 19, TypeScript, tRPC, TanStack Query, Prisma on Postgres, Tailwind, and Radix: a feature-first module structure, a Storybook-documented component library with accessibility checks, Vitest unit coverage, and Playwright end-to-end suites.

Held formal technical-direction authority for WeWork's global mobile product and platform organization, including Android, inside an approximately 1,000-person technology group across four hubs. Owned the Android platform team, set architecture, engineering standards, and investment priorities, held final technical accountability for execution, and partnered with Engineering Managers, executives, product, design, and research on organization and portfolio strategy.

• Directed the mobile platform behind WeWork's ground-up member-app rebuild, enabling discovery, booking, payment, and workspace entry across a network that reached 779 locations in 39 countries and 682,000 physical members.
• Helped scale the Access product line to roughly 70,000 memberships and $200M in annualized revenue, growing On Demand from a 12-building New York pilot to 330+ locations.
• Owned the Android platform team chartered to build observability and performance tooling for the wider mobile organization, running WeWork's member app at a 99.8% crash-free rate while enabling product teams to ship faster with consistent quality.
• Unified 15+ teams running separate design systems into one shared mobile architecture (Kotlin/MVI, reusable components, architecture reviews, engineering guilds) driven by a single server-side source of truth, raising developer productivity and release consistency across the organization.
• Led the mobile-credentials program from hardware R&D through a production-ready integration and global rollout handoff across an initial 528-location, 29-country footprint, addressing approximately $50M in physical-card cost exposure; built Arduino test rigs, reverse-engineered vendor SDKs, and contributed fixes upstream.
• Led technical onboarding of Prolific Interactive's engineering teams in New York, San Francisco, and Durham onto that operating model after WeWork acquired the 200+ person product agency.
• Built WeWork's mobile hiring system: an interviewer-certification program that grew certified mobile interviewers from 4 to 25+, a take-home assessment adopted company-wide that re-qualified previously rejected candidates (several became some of the organization's strongest engineers), and interviewing guidelines that became standing policy.

Joined as a product engineer and grew into Lead Product Engineer, Mobile, the most senior engineer on mobile web and native apps, building nearly every matching, messaging, monetization, and identity feature OkCupid shipped to mobile for roughly 25M people per year at peak.

• Owned the rebuild of a largely web-based mobile app as a native product on a server-driven thin-client architecture that improved load time 10x, kept A/B-through-Z experimentation fast with logic and observability centralized server-side, served approximately 4.2M people per month, and held a 99.8% crash-free rate.
• Built the mobile web and native implementation of the experiment that retired profile-view tracking and shifted discovery toward mutual interest, lifting new-user retention and producing 10% more mutual likes (mutual matches were already 4x more likely to converse).
• Engineered the mobile side of a year-long messaging and ranking redesign, validated through a four-week test across 18 regions and nearly 1M accounts before open messaging was replaced with mutual-interest delivery.
• Built the mobile implementation of Boost, one of dating apps' first pay-to-prioritize monetization features, which helped shift OkCupid from roughly half ad-supported revenue to primarily premium offerings and was later adopted across Tinder, Hinge, and Match.
• Led OkCupid's 2014 identity expansion from 2 gender and 3 orientation choices to 22 gender identities and 13 sexual orientations across profile data, preferences, matching, filtering, and discovery; coverage reached PBS NewsHour and Time.
• Partnered with senior product and data science leadership on multivariate experiments across matching, profiles, messaging, ranking, and engagement, owning the mobile experimentation surface.
• Contributed to automated fraud detection, enforcement, reporting, and human-moderation workflows that protected users and messaging quality while removing approximately 500,000 fraudulent accounts per month.
• Published the thin-client architecture on the OkCupid Tech Blog ("Engineering Smart && Building Dumb") and presented it at QCon San Francisco 2018 (InfoQ: "Engineering Dumb: Modern Mobile Thin Clients"), Google, Facebook's Mobile Consortium, and Droidcon Singapore, and exchanged consumer-product practices with Tinder and Hinge.
• Led OkCupid's first diversity-hiring pipeline with recruiting and C4Q (Coalition for Queens), designing group interviews that raised the mobile engineering pass rate from roughly 5-10% to almost 40% and produced the division's first female and more racially diverse engineering hires.

Created the company's first mobile companion to its association-management SaaS platform and a reusable cross-platform framework adopted by more than 70 associations.

Co-founded a software consultancy and led a small agile team from client discovery through delivery of web applications and product prototypes.

Led full-stack delivery of a resume-indexing and candidate-fit platform for government recruiting workflows and managed a small intern team through production deployment.