Descricao da vaga
About LumiMeds LumiMeds is a fast-growing U.S.-based telehealth startup focused on weight management and metabolic health.
We're building next-generation e-commerce and clinical infrastructure from the ground up.
We move fast, operate lean, and expect high ownership from every team member.
What We're Building An AI-accelerated clinical engine, high-converting e-commerce storefront, consumer mobile app, and intake infrastructure.
We don't scale by adding headcount — we scale by multiplying leverage through better systems and AI.
Our engineers already orchestrate Claude agent teams in production.
The Role A player/coach role for an engineer-turned-manager serious about AI-native engineering.
You'll lead 4–8 full-stack engineers and serve as the team's resident expert in designing and running Claude agent teams. ~40% coding and agent pipelines, ~60% managing and raising the team's AI literacy.
What Makes This Role Different Most engineering managers run teams of humans.
You will run a team of humans and a fleet of agents — and your ability to design, direct, and verify agent work is as important as your ability to manage people.
We are not looking for someone who has used Claude or experimented with agents.
We need someone who has built real agent pipelines under production pressure — who knows exactly where they break, how to recover, and how to design for reliability at scale.
This is the core of the role.
Break complex engineering initiatives into agent-executable subtasks with crisp acceptance criteria Design multi-agent pipelines — parallel subagents for feature branches, test generation, code review, and documentation — and stitch their outputs into production-ready deliverables Set the team's standards for when to use agents, how to prompt them effectively, how to verify their outputs, and when to override them Treat agent output as team output — you are accountable for everything the agents on your team produce Continuously raise the ceiling: as the models improve, you update the playbook What You'll Do Lead and grow a high-performing engineering team of 4–8.
Hire, onboard, coach, and develop engineers with direct feedback and clear expectations.
Design and operate Claude agent pipelines — parallel feature development, test generation, code review, documentation, and spec-to-implementation workflows.
Stay in the code.
Write production code, review PRs with technical depth, and debug hard problems.
AI augments your output — it doesn't replace your judgment.
Set the AI velocity standard.
Define how the team uses Cursor, Claude Code, and agent pipelines.
Push the frontier.
Own delivery end to end.
Sprint planning, blocker resolution, cross-functional coordination across product, clinical, and infrastructure.
Build and own consumer app and e-commerce systems — storefronts, checkout, subscriptions, billing, and user behavior tracking infrastructure.
Build and own A/B testing infrastructure — feature flags, experiment assignment, statistical tracking, and results dashboards.
Collaborate cross-functionally with Product, Clinical, and Ops as the technical decision-maker in roadmap conversations.
What You'll Own Domain Scope Engineering Team 4–8 full-stack engineers — hiring, performance, growth Agent Team Operations Claude agent pipelines for parallelized engineering work Delivery Sprint execution, unblocking, cross-functional coordination Technical Standards Code quality, PR norms, AI tooling standards, observability Core Platform Clinical engine, e-commerce, subscription billing, mobile backend A/B Testing Platform Experimentation infrastructure, feature flags, statistical tracking AI Infrastructure LLM integration patterns, agent orchestration architecture Required Experience Engineering Leadership: 6+ years of software engineering experience, including 2+ years in a lead or management role Hands-on experience with Node.js / TypeScript backends and Next.js / React frontends — you can read, write, and review production code at a senior level Strong database fundamentals: PostgreSQL (schema design, migrations, query optimization), Redis AI-Native Engineering (Non-negotiable): Claude Agent SDK: Demonstrated experience building and orchestrating multi-agent pipelines — decomposing tasks, defining subagent roles, managing context handoffs, validating agent output LLM Integration: Production experience integrating LLMs into real systems — streaming, tool use, structured outputs, prompt engineering AI Dev Tooling: Daily use of Claude Code, Cursor, or equivalent.
You have built workflows around these tools, not just used them ad hoc You can articulate — with specificity — how agent orchestration changes what a small engineering team can ship Experimentation & A/B Testing: Proven experience designing and building web A/B testing platforms from the ground up — not just using third-party tools, but owning the infrastructure Deep understanding of experiment design: randomization, assignment consistency, statistical power, holdout groups, and avoiding novelty bias Experience running experiments across high-traffic consumer funnels (checkout, onboarding, pricing, landing pages) Familiarity with feature flag systems (LaunchDarkly, Statsig, homegrown) and experimentation analytics pipelines Consumer Product & Tracking: Hands-on experience building consumer apps and e-commerce platforms end to end — storefronts, checkout, subscriptions, billing Built user behavior tracking infrastructure: event schemas, analytics pipelines, conversion funnels, retention analysis Familiarity with tools like Segment, Mixpanel, Amplitude, or equivalent homegrown tracking systems Systems & Delivery: Experience running engineering sprints, managing dependencies, and owning delivery timelines Ability to write engineering specs that AI coding agents and engineers can execute with minimal back-and-forth Familiarity with AWS (EC2, RDS, Lambda, S3), Vercel, GitHub Actions, and CI/CD pipelines Compliance: Working knowledge of HIPAA/SOC2 requirements — you understand how compliance shapes architecture decisions English: Fluent written and spoken English — all team communication is async in English (Slack, PRs, specs, docs) Candidate Qualifications The player/coach instinct: You want to manage, but you're not ready to stop building.
You think leaving code entirely would make you a worse manager.
AI-native, provably so: You can show, concretely, how agent orchestration has changed your own output — examples, numbers, or a portfolio.
Not just familiarity — results.
High standards for output quality: AI-generated code is your code.
You are not the manager who merges anything that compiles.
You have a verification practice.
Direct communicator: You give feedback clearly, make decisions without excessive consensus-building, and disagree with product or leadership when the technical reality demands it.
High agency: You identify problems, propose solutions, and execute — you don't wait to be managed.
Nice to Have Experience in telehealth, DTC health, or a regulated healthcare environment You've built an internal agent framework or tooling layer that other engineers on your team used Shipped a consumer mobile app with measurable retention and a backend you owned Background in distributed systems or real-time infrastructure (WebRTC, event-driven architectures) You've written a post, given a talk, or built something in public about AI-augmented engineering Why LumiMeds AI as infrastructure, not a feature.
We've already rewired how we build around AI.
You won't be evangelizing something new — you'll be operating at the frontier with a team that's already bought in.
Real technical complexity.
Clinical state machines, real-time patient-provider flows, high-stakes billing, HIPAA.
The problems are hard because the domain is hard.
Small team, enormous leverage.
You won't manage through layers.
Your decisions show up in production the same week.
Direct impact.
The systems you build affect patient outcomes.
That's not a cliché here — it's the constraint that makes the work matter.