Software Engineer, AI

About the role

Bilt is the membership for where you live. It started as a way to reward people on their largest expense (rent) and has grown into an ecosystem connecting members, property managers, and local merchants across travel, dining, fitness, and everyday spend. The Concierge role runs in the Bilt app, on the web, and over SMS, iMessage, and email, handling tens of thousands of conversations a day with monthly spikes. You will build and run the backend behind all of it.

Responsibilities

• Design, implement, and operate backend services for the Concierge agent platform end‑to‑end (agent capabilities, tools, workflows, prompts, integrations).
• Build and maintain the agent runtime (multi‑agent orchestration, context and memory, prompt caching, routing across frontier and open‑weight models).
• Ensure performance (p50/p99 latency, bursty traffic handling, cost per conversation) and develop eval harnesses, LLM judges, and conversation replay.
• Create assistants for merchant staff that leverage business‑specific data and browser agents that act where no API exists.
• Develop proactive engagement features (briefings, follow‑ups, outreach) for members and merchants.
• Build the conversation platform: a unified inbox for agents, merchant staff, and Bilt support across SMS, iMessage, email, web, and in‑app chat.
• Implement real‑time messaging with streaming replies, attachments, ordering, idempotent delivery, carrier and iMessage limits, and human hand‑off.
• Design authentication and fine‑grained authorization so the agent only accesses permitted data and actions.
• Contribute to an AI‑native engineering culture: writing, reviewing, debugging, and operating code with AI agents, sharing context via playbooks, and owning features from design through rollout and on‑call.

Requirements

• 4+ years of experience building and operating backend services in production.
• Proficiency in at least one backend language; primary stack includes Python (FastAPI), Java (Quarkus), and Go on Google Cloud.
• Solid distributed systems fundamentals (async processing, pub/sub, idempotency, retries, caching) and ability to debug across services using logs, traces, and metrics.
• Hands‑on experience with LLMs (at work or personal projects) and curiosity about making them reliable.
• Product judgment: ability to diagnose whether a bad conversation stems from the prompt, a tool, the data, or a service.
• Daily use of AI coding agents and good judgment about when to trust their output.
• Clear writing skills for design docs, PR descriptions, and prompts given to agents.

Qualifications

• Shipped an LLM‑powered feature to real users and can explain failures and measurements.
• Experience with agent frameworks (Google ADK, LangGraph, or similar), tool calling, evals, LLM‑as‑judge.
• Model serving and inference expertise: provider routing, open‑weight models, latency and cost tuning.
• Background in messaging or real‑time infrastructure.
• Browser automation or computer‑use agents experience.
• Knowledge of identity and access (OAuth/OIDC, token exchange, fine‑grained authorization).

Skills

• Languages: Python, Java, Go
• Frameworks: FastAPI, Quarkus
• Cloud: Google Cloud
• LLM tooling: prompt caching, eval harnesses, LLM judges
• Messaging: realtime streaming, idempotent delivery
• Auth: OAuth/OIDC, fine‑grained authorization
• Observability: logs, traces, metrics, feature flags

Benefits

• Competitive salary with equity and annual performance‑based bonus.
• Health insurance for you and your loved ones from day one, including One Medical membership, wellness stipends, family programs.
• 401(k) plan with company match.
• Commuter FSAs to reduce transit costs.
• Unlimited PTO.
• Exclusive employee‑only Bilt points.
• Team events to foster human connection.

Pay

• Salary range: $170,000 – $300,000 (based on years of experience/level), plus equity and annual performance‑based bonus.