Founding Member of Technical Staff, AI Infrastructure

Why this work matters

Touchdown starts with people doing real work: teach the team, build one useful AI system, manage what runs, and measure what still breaks. We follow repeated limits through software, inference, kernels, memory, hardware, materials, and manufacturing research only when the evidence earns the next layer. The ambition is broad. The work is always bounded by a user, a test, and a receipt.

Overview

This is a broad founding engineering role for a senior builder who can start with a person doing real work, define what a useful result means, and build the system that produces it. You will move between product, backend services, agent workflows, private evaluation, evidence, managed inference, and infrastructure without losing the customer outcome. This is not a narrow backend role and not a consulting role. The whole point is to own ambiguous technical problems end to end and follow the actual bottleneck rather than protect a narrow boundary.

What you will own

• Work directly from a repeated customer job, its acceptance criteria, failure modes, and operating constraints.
• Design and ship the full application path: context, tools, state, agent control, APIs, data, evaluation, and deployment.
• Decide when a problem belongs in product code, workflow design, model selection, retrieval, inference, or lower-level infrastructure.
• Build replayable tests that preserve accepted, rejected, and invalid outcomes instead of optimizing a demo.
• Connect technical changes to quality, latency, reliability, cost, and the ability of the customer's team to operate the system.
• Create durable code, documentation, runbooks, and receipts that another engineer can review, reproduce, and extend.

Technical territory

Agent and AI-native engineering; context and tool contracts; retrieval and state; evaluation and human review; backend and product systems; model and provider qualification; routing, caching, inference, observability, reliability, security, and cost. You do not need to be the deepest person in every layer, but you must know how to find the owner, evidence, and next test when the bottleneck moves.

Representative outputs

• A versioned workflow and acceptance contract that names the user, inputs, tools, reviewer, accepted result, failure states, and operating constraints.
• A working system with explicit state, evaluation, retry, fallback, security, deployment, and rollback behavior.
• A joined evidence packet connecting user outcomes to application, model, runtime, infrastructure, and cost events.
• An operator handoff with code, tests, runbooks, unresolved risks, and the next controlled experiment.

What success looks like

• A customer or internal operator can use the system independently on the defined job.
• The acceptance test, failure states, and rollback path are visible and reproducible.
• The team can explain where quality, latency, reliability, and cost come from.
• What was learned survives a model, provider, or infrastructure change.

What you bring

• Strong software or systems engineering judgment and a record of shipping real products or infrastructure.
• Ability to turn an underspecified workflow into requirements, interfaces, tests, and an executable plan.
• Ability to learn unfamiliar layers quickly while keeping ownership of correctness and operational risk.
• Clear written communication and technical decisions tied to code, traces, measurements, or user outcomes.
• Experience in Python, TypeScript, Go, Rust, C++, or comparable production languages.
• Comfort working directly with customers, researchers, and engineers at different levels of the stack.

Helpful experience

• Agent systems or ML infrastructure
• Evaluation, observability, or developer tools
• Inference or distributed systems
• Security and privacy boundaries
• Early-stage or zero-to-one work

How the role works

• Full-time role.
• San Francisco / Bay Area preferred. Remote within the United States may be considered for the right person.
• Scope, start date, and employment details are discussed during the process.