SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Fello is a profitable, hyper-growth, VC-backed B2B SaaS company building an agentic real estate platform. The platform combines data intelligence, marketing automation, and conversational AI to help real estate teams engage smarter and scale faster, powered by Felix, an AI teammate that automates follow-up across calls, texts, and emails.
You will own the data architecture across all GTM applications, including schema design, object modeling, access patterns, and the path from raw source to what agents and users read. Key responsibilities include:
- Build and enforce a data dictionary where every entity and metric (account health, sentiment score, risk, lifecycle stage) has one canonical, versioned, deterministic definition reused across all apps.
- Design a ledger system for scale, recording metadata history (who did what, where leads came from, what agents changed) with decisions on hot vs. cold storage and queryability for debugging and audit.
- Bridge R&D and production systems by re-architecting fast-built systems to run in production, serving as the quality gate.
- Design for scale proactively: caching layers, indexing, streaming/queueing instead of database hammering, and event-driven agent runs. Monitor latency, query cost, and token spend.
- Give agents proper data access through SQL tools, MCP servers, and endpoints so they fetch exactly what they need, reducing cost and hallucination.
- Own access and security: RBAC, per-team/per-app views, tenant isolation, PII handling, and secrets management.
- Build and harden agents that execute real workflows with guardrails: prompt injection defense, human-in-the-loop for customer-facing work, validated and cited output, and evals that catch regressions.
- Make deterministic-vs-probabilistic calls on every system and ship whichever wins.
- Set the engineering bar: architecture docs before implementation, reviews that teach, reusable patterns, and runbooks.
- Mentor a fast, young, AI-native team that ships quickly with coding agents, ensuring they understand the systems they generate and think in objects, schemas, and contracts.
- Work across multiple products (CS, sales, marketing, metrics) simultaneously while keeping architecture coherent.
You are a software engineer first who spent years building and scaling production systems before LLMs, with significant time on database, data platform, or infrastructure problems. You've gone deep on AI: agents, tool use, structured outputs, retrieval, memory. You know where a schema, rule, or small model beats a prompt, and you don't ship coding-agent output you can't explain. You're opinionated about structure and comfortable being the person who slows things down for the right reasons. You want to stay deep in technical work rather than move into pure management, are happy juggling multiple products without a handed roadmap, and give and take blunt technical feedback.
REQUIREMENTS:
- 7+ years of software engineering, mostly pre-LLM, building systems from the ground up.
- Deep data architecture experience: structured and unstructured data modeling, warehouse and data lake design, ETL/ELT pipelines, schemas that scale.
- Real distributed backend experience: services carrying significant load with caching, streaming (Kafka, Pub/Sub or similar), and queueing.
- Hands-on cloud depth, ideally AWS, knowing which services to pick (Athena, S3 tiers, RDS/Postgres) and how to tune for speed and cost.
- Strong Postgres skills including performance tuning, indexing, and query design (Supabase is a plus).
- Hands-on production experience with modern LLMs (Anthropic, OpenAI, open models): agents, copilots, or workflows shipped and maintained.
- Reliability instincts: structured prompting, output validation, evals, observability.
- Security posture: authz, RBAC, secrets, PII, tenant isolation, SOC 2 awareness for B2B SaaS.
- Experience leading or mentoring engineers with a track record of raising team floor.
Nice to have: time on database/data platform/infra team at scale; designed monorepo or internal platform; chat-first assistants taking real actions; small language models or decision models replacing LLM calls on high-volume tasks; agent observability and evals in production; vector search and retrieval at scale; CRM and revenue ops data models; real estate or proptech context.
