Software Engineer (Core Systems)

Help us push the frontier of autonomous AI agents.

About Dialtone

Dialtone is building the platform layer for enterprise-grade agentic AI. Our stack spans real-time orchestration, retrieval-augmented reasoning, dynamic tool calling, and human-in-the-loop oversight - all delivered as production SaaS. Enterprises rely on Dialtone Agents to automate thousands of hours of cross-system work each month.

We’re not just gluing APIs together; we’re inventing deeply technical infrastructure that makes large-language-model reasoning reliable, observable, and secure at scale.

What you’ll build

• Low-latency inference & routing
  Spin up and optimize distributed LLM endpoints (OpenAI, Anthropic, Ollama) with real-time model selection and cost introspection.
• Agent planning & memory engines
  Design hierarchical planners, episodic memory stores, and long-horizon task schedulers that let agents reason over weeks - not seconds.
• Retrieval & grounding pipelines
  Blend vector search (Qdrant / PGVector) with structured SQL/Graph data to keep agents factual and auditable.
• Dynamic tool & plugin framework
  Sandbox arbitrary API calls so agents can safely manipulate external systems inside customer VPCs.
• Human-in-the-loop UX
  Craft real-time dashboards where operators can invoke agents, supervise, intervene, and reinforce agents - closing the RLHF loop.
• Rock-solid platform primitives
  Harden auth (SAML/SCIM), multitenant state machines, and event queues that sustain >99.99 % uptime SLAs.

You might be a fit if you…

• Have 3-10 yrs building production software across front- and back-end layers, with a systems-level mindset.
• Write idiomatic, tested TypeScript / React and Node.js / Python - and know when to drop to Go or Rust for performance.
• Have grokked distributed systems: you’ve scaled queues, caches, or real-time data pipelines into the billions.
• Are energized by LLM ops (prompt engineering, eval harnesses, fine-tuning, gradient caching).
• Love turning research papers into shippable product features.
• Default to ownership: you spot a problem, propose a fix, ship, and measure.

Extra credit

• Published or contributed to open-source LLM / agentic frameworks.
• Familiarity with ReAct, Chain-of-Thought, or DSPy patterns in production.
• Proficiency in CUDA, Triton, or other GPU-level optimizations.

Why Dialtone?

• Front-row seat to AGI-era systems - work on problems few teams have solved.
• High leverage - small team, massive ownership of architecture and roadmap.
• Competitive comp + meaningful equity - you win when the company wins.
• Work style - Onsite in Mountain View, CA.
• Hardware stipend • full health/vision/dental • 401(k) match • unlimited PTO.

We move fast: you’ll meet the founders within a week.

Join us to architect the future of agentic automation - one deeply technical breakthrough at a time. 🚀