We’re rebuilding psychiatry from the ground up—AI-native, insurance-covered, and engineered to scale.

Legion Health is a high-quality psychiatry network where patients get care from licensed clinicians, fast and affordably. Under the hood, our LLM agents and orchestration layer automate operations, reduce admin, and enable truly great care.

Most startups bolt AI onto broken workflows. We’re replacing the workflows—and the system underneath.

If you’ve ever wanted to use AI to actually fix something that matters, this is your shot.

Part‑time (≈20 hrs/week) • 3‑month project • Remote (US‑friendly hours)

🧠 About Legion Health

Legion Health is on a mission to deliver the best psychiatric care in the world—and to do it at scale.

We're building a full-stack AI-native psychiatry network—our clinicians provide high-quality care directly to patients, and AI agents automate all parts of our care operations (e.g., scheduling, risk analysis, billing, etc.). Our vision? A patient experience that’s 10X better, more affordable, and higher-margin than ever before.

Today, we're turning weeks-long wait times into same-day appointments, crafting seamless digital experiences in a broken healthcare system, and scaling a company that patients genuinely trust.

⚡ Why This Role Matters

Legion Health’s data lives in half‑automated dashboards, ops spreadsheets, and ad‑platform exports. We need someone to wire everything together, instrument every click and server outcome, and make growth metrics self‑serve for the entire team.

📦 What You’ll Deliver

1 – Centralized Growth Metrics & Dashboards (Required)

• Pipe data from GA4, ops scorecards, billing, EMR, and provider‑availability feeds into one source of truth in PostHog/Segment.
• Centralize and automate the collection of the KPIs that matter for our insurance‑based healthcare service:
  • Paid and organic CAC, plus CAC per retained patient
  • Step‑by‑step onboarding conversion (landing page start → eligibility complete → benefits verified → intake finished → first appointment booked and shown)
  • Real‑time provider availability and idle time
  • Time‑to‑intake and time‑to‑first visit
  • Patient retention, weekly LTV, churn, and net margin per visit
  • Claim‑denial rates and gross‑to‑net reimbursement
  • etc.
• Automate dashboards, Slack/email digests, and threshold alerts (e.g., CAC spikes, idle‑provider alerts) for real‑time decision‑making.
• Keep all metric definitions in a living data dictionary so finance, growth, and clinical ops speak the same language.

2 – End‑to‑End Analytics Stack (Future, if Successful in #1)

• Fully implement PostHog, Segment, and Google Analytics 4 for complete funnel and behavioral insights.
• Instrument every product event with both client‑side “click” and server‑side “success” calls; build a structured tracking plan that documents event names, purpose, properties, and collection points.
• Deploy Segment for clean, schema‑driven event collection and bi‑directional integrations with ads, email, and data warehouse.
• Validate that our event‑driven architecture flows seamlessly from first click to patient conversion and downstream ops automations.

3 – Democratized Data Access (Future, if Successful in #1)

• Stand up BigQuery (or Snowflake) as the warehouse of record; schedule Segment/PostHog exports.
• Layer a BI tool such as Mode or Metabase for ad‑hoc SQL analysis.
• Build a shared library of saved queries and dashboards for recurring metrics (e.g., funnel drop‑off by payer, LTV by lead source).
• Run hands‑on workshops so ops, product, and marketing teammates can adapt basic SQL and pull their own insights without waiting on you.

4 – Reporting & Knowledge Transfer (Required)

• Replace manual daily roll‑ups with self‑updating views accessible company‑wide.
• Deliver concise weekly memos summarizing data quality, new instrumentation, and experiment read‑outs.
• Document the full event taxonomy, data model, and dashboard logic; hand off maintenance playbooks to future data or growth hires.

🏆 You’ll Be Successful If You…

• Automate and centralize every spreadsheet KPI and standalone tool metric into dashboards the team actually uses.
• Cut reporting lag to near‑zero—finance, ops, and growth can answer ad‑hoc questions in minutes, not days.
• Surface anomalies before humans do—threshold alerts flag CAC spikes, idle providers, or denial‑rate jumps the same day they happen.
• Standardize our language of metrics—finance, growth, and clinical ops all quote the same CAC, LTV, and churn numbers without debate.
• Shrink manual data work by 80 %—recurring exports, VLOOKUPs, and copy‑pastes vanish under your pipelines or scripts.

🧰 Must‑Have Skills

• Analytics ingestion & modeling: Hands-on experience piping data from GA4, ops scorecards, billing systems, EMR exports, and provider-availability feeds into a modern analytics platform (e.g., PostHog, Mixpanel, Amplitude, etc.) as a single source of truth.
• KPI centralization & automation: Proven ability to define, calculate, and automate key insurance-based metrics (CAC, onboard-to-visit funnels, provider availability, time-to-intake/visit, retention, LTV, churn, denial rates, net margin, etc.).
• Dashboard & alerting: Skilled at building self-updating PostHog dashboards, Slack/email digests, and threshold alerts (e.g., CAC spikes, idle-provider warnings) for real-time decision-making.
• Data dictionary governance: Experience creating and maintaining a living data dictionary so finance, growth, and clinical ops teams share one consistent vocabulary.

🎁 Nice‑to‑Have Skills

• Alternate event pipelines: Experience with Segment, RudderStack, or similar for schema-driven event collection
• Warehouse & BI tools: Hands-on with BigQuery or Snowflake setup/maintenance and BI platforms (Mode, Metabase, Looker) for ad-hoc analysis.
• Advanced SQL analytics: Comfort writing complex joins, window functions, and cohort/LTV analyses across product, revenue, and ops tables.
• JS/TS tagging & schema design: Fluency in JavaScript/TypeScript client- and server-side event instrumentation and data schema modeling.
• Data-ops communication: Proven ability to teach non-technical teammates to self-serve metrics—clear documentation, workshops, and playbooks.
• HIPAA-adjacent analytics: Familiarity with telehealth or healthcare-compliant data stacks and workflows.
• Experimentation & attribution: Knowledge of attribution modeling, A/B testing frameworks, or PostHog Experiments API.

💰 Engagement Details

• Time: ~20 hrs/week for X months.
• Compensation: Competitive hourly or project rate.
• Resources: Budget for any tooling you recommend; direct access to engineering, growth, and ops leads.

Technology at Legion

We are revolutionizing telepsychiatry through a core AI infrastructure that enables a vastly improved patient experience and quality of mental health care. We are actively improving our services in ingesting, chunking, embedding, caching, and searching to achieve this goal. Our technical goals include:

• Enhanced processing: Implement robust data structures and caching mechanisms to handle increasing volumes of language data spanning every facet of patients’ care while minimizing latency and maximizing system responsiveness.
• System scalability: Ensure our systems can scale effectively to support growth from one million to hundreds of millions of data points.
• Securing sensitive data: Uphold the highest standards of data security and privacy in compliance with HIPAA guidelines.

##Stack TypeScript, Node.js, PostgreSQL, Redis, Supabase, Azure, Google Cloud, AWS, LangChain, PostHog, Retool