Solutions Engineer (HealthTech)
Level: Senior/Staff
Type: Full-time
The Company
We're partnering with an AI startup building an AI operating system for healthcare clinics in Latin America.
Its AI agent answers patients on WhatsApp, books and follows up on appointments, and hands conversations over to the clinic's team when a person should step in. As it works, the platform turns every interaction into structured patient context, so each doctor a patient sees starts with the full story.
The company's mission is to make personalized care available around the clock, starting with Latin America. This is an opportunity to help define how AI agents support patients and healthcare teams.
The Role
This is a forward-deployed engineering role for AI agents in healthcare. You'll work where the model meets the patient: inside real clinic workflows, reading real conversations and deciding how an agent should behave when the stakes are someone's health and a clinic's business.
You'll own the agent in production, from the prompt to the runtime to the evaluation that proves a change worked. The best people in this seat don't stop at fixing the hard case in front of them. They turn it into a capability every clinic gets, so the next hundred clinics launch faster than the last ten.
You'll join Solutions Engineering, reporting to the Head of Solutions Engineering, and work closely with Platform, Product, and Customer Success. What you ship reaches real patients within weeks.
What You'll Do
Design how the agent talks to patients. Decide when it should ask, wait, reassure, or hand off to a person, then encode those decisions in prompts, state machines, and guardrails that hold up across specialties and thousands of conversations.
Build the loop that makes agents better every week. Create evaluations drawn from real conversations, traces that explain failures, and safe rollouts, so every prompt, model, or configuration change ships on evidence.
Turn edge cases into platform capabilities. A fix for one clinic becomes a reusable workflow, template, or no-code control for all of them. Your work compounds with every deployment.
Make launching a clinic self-serve. Shape the onboarding schemas, validators, and guided setup that let Customer Success deploy a new clinic without an engineer in the room.
Take new capabilities from idea to production. Scheduling, follow-ups, reminders, and whatever clinics need next. You'll help decide what the agent learns to do, connect it to the EHRs and systems clinics already use, and ship it in weeks.
Choose the right model for every job. Evaluate and swap models across providers, balancing quality, latency, and cost, and prove the tradeoff with data.
Work agent-first. The team builds with coding agents every day, with shared skills for trace analysis, prompt tuning, evaluations, and clinic configuration. You'll use them, extend them, and help define how a small team gets more done with AI.
What Success Looks Like
Within your first months, real patient conversations are measurably better because of changes you shipped, and you can show the evaluation that proves it.
A problem one clinic encountered becomes a capability every clinic has.
Model upgrades ship without regressions because your evaluations caught the issues first.
Customer Success launches new clinics without engineering help.
The agent gets measurably better every week through a system that doesn't depend on constant manual supervision.
You Are
Fluent in Spanish, native or near-native, with professional English, and comfortable reading real patient transcripts.
Obsessed with the conversation. Tone, pacing, trust, and when to escalate matter to you, and you back your opinions with data, evaluations, and traces.
A builder who ships to real users. You want your work in production, and you close the loop with the people using it.
A systems thinker. You'd rather solve a class of problems once than the same problem twice.
Pragmatic. You deliver the 80/20 fast, polish where it matters, and leave a clean path forward.
An owner. Proactive, decisive, and comfortable with ambiguity.
AI-native, not AI-dependent. Coding agents make you faster, and you can walk anyone through the design and tradeoffs of everything you shipped.
Minimum Qualifications
5+ years in implementation engineering, solutions engineering, or product engineering with customer ownership.
Strong Python; fluency with HTTP and webhooks, including authentication, signatures, idempotency, and retries; JSON and schema validation.
Hands-on experience building agentic systems: writing and maintaining high-performance prompts, and iterating on them with evaluations and traces in event-driven architectures.
Data mapping and integration experience; ability to design adapters and validators.
Production instincts around observability, operational playbooks, and safe rollouts.
Nice to Have
Healthcare or EHR exposure, familiarity with FHIR, or firsthand experience running clinic operations.
WhatsApp Cloud API, Chatwoot, Twilio, or similar messaging platforms.
Experience building admin tools, wizards, or DSLs for non-technical users.
SQL and comfort analyzing conversation data at scale.
Portuguese proficiency.
Tech Stack
Languages: Python, TypeScript.
LLMs: OpenAI and Anthropic models on Azure, through an internal AI gateway.
Healthcare & Messaging: FHIR (Medplum), WhatsApp Cloud API, Chatwoot.
Databases & Storage: PostgreSQL, Redis, S3, GCS.
Data & BI: BigQuery, Dataform, Metabase.
Cloud & Infrastructure: AWS, GCP, Azure, Vercel, Kubernetes, Terraform.
APIs & Middleware: FastAPI.
Observability & Evaluations: Arize Phoenix, deepeval, Prometheus, Grafana.
CI/CD: GitHub Actions.
AI Development: Claude Code, Codex.
Collaboration: Slack, Notion, Attio, Lucidchart, Linear.
Working Model
Fully remote within Latin America.