Future - Staff Applied AI Engineer
At a glance
Senior AI Engineering role at deCircle. Remote, United States · no salary stated · we estimate $205,000–$277,500.
deCircle has not published a salary for this role. The figure below is our estimate, not theirs.
aimlroles summary, based on the employer's posting.
What you'll do
- Build and release production AI agents for Health Coach, including tool use, structured responses and parallel API calls
- Create streaming features for coach matching and health guidance, with human review built into the flow
- Run the agent feedback cycle: prototype, assess with Langfuse rubrics, deploy, inspect traces and refine
- Reduce latency, cost and failure rates through caching, token controls, retries and better observability
- Develop Pydantic-validated integrations with the EPD pod
What you bring
- You have shipped LLM agents into production and bring four to ten years of relevant experience
- You can build in Python and async Python, using LangChain or LangGraph and Pydantic
- You have worked with Langfuse and AWS or Bedrock in applied AI systems
Who this fits
This role suits an independent engineer comfortable building in a beta-stage environment. It is fully remote across US time zones, with optional anchor weeks at the Los Angeles headquarters. Applicants must be US citizens or Green Card holders.
From the employer
Future (future.co): personal fitness/health coaching; building the AI foundation of a new Health Coach product. $110M raised; board includes Tom Brady, Eddie Cue, Adam Bain.
Comp: $250K-$300K base (SF rates for all US) + competitive equity. Location: Fully remote, US time zones; optional anchor weeks at Brentwood LA HQ. Visa: Not open to visas (US citizen / Green Card).
Responsibilities
Build and ship production AI agents for Health Coach: tool-calling LLMs, structured output, parallel API orchestration, streaming
AI features: coach matching, clinical-leaning guidance (sleep, nutrition, fitness), human in the loop
Own the loop: prototype, Langfuse evals/rubrics, deploy, monitor traces, iterate
Improve reliability/latency/cost: prompt caching, token budgets, retries, observability
Write agent tools (Pydantic-validated integrations) in an EPD pod
Tech stack: Python, Async Python, LangChain, LangGraph, LLMs, Pydantic, JSON Schema, SSE, Webhooks, Langfuse, AWS, AWS Bedrock, OpenTelemetry, Datadog, Terraform, CDK
Source: https://www.paraform.com/share/future/cmur5lxlk000g06nis011p9o1
4-10 yrs; shipped LLM agents to production
Strong big-tech background, ideally moved to startups or founded a company
Python/async, LangChain/LangGraph, Pydantic, Langfuse, AWS/Bedrock
Independent in a beta-stage environment; US citizen/GC