Foundry Process & Equipment Lead
Choose the first qualified process path, name its owners, and reject work that cannot meet safety, product, or cost gates.
Careers · AI Infrastructure and Foundry
Touchdown owns the customer relationship for scoped AI infrastructure work while engineering an American fab and foundry in parallel. We are hiring across software, systems, advanced packaging, lithography, process integration, equipment, test, quality, and operations. The fab program is in engineering; no production fab is operating today.
These are focused application lanes, not rigid boxes. We care about evidence, learning speed, ownership, and the ability to connect your layer to the full system.
/01 · Foundry and fab
We are engineering a modular U.S. foundry path, starting with a high-mix advanced packaging direction. These are application lanes for the work and evidence needed to qualify a route. We do not operate a production fab today. Read the Foundry hiring board →
Choose the first qualified process path, name its owners, and reject work that cannot meet safety, product, or cost gates.
Turn separate process steps into a controlled route with measurable handoffs, holds, and yield learning.
Define an advanced package from known-good die through assembly, test, reliability, and customer acceptance.
Qualify a practical patterning route for a named package layer, with overlay, dimensions, defects, and cost under control.
Qualify films and conductors against electrical, adhesion, uniformity, contamination, and reliability limits.
Define selective removal and cleaning steps that protect the workpiece and leave measurable surfaces for the next process.
Choose and qualify assembly joins, alignment, inspection, and rework limits for a real package.
Make each process claim measurable with calibrated methods, sampling, uncertainty, and independent reject rules.
Build test coverage and limits that turn a package or device into an accepted electrical result.
Trace yield loss and reliability failures to measured mechanisms, then verify corrective action against a baseline.
Evaluate and qualify process equipment, then build safe maintenance and recovery paths for an authorized line.
Test whether a candidate site can safely support the process, equipment, people, and waste streams.
Make every lot, recipe, material, tool state, measurement, hold, and release traceable.
Set acceptance, hold, corrective-action, and supplier controls before a product or process can be qualified.
Plan qualified lots against real capacity, holds, yield, materials, and customer commitments.
Compare buying, partnering, and building against the same product specification, cost, rights, and risk.
Turn a buyer's workpiece into a written qualification test, partner baseline, and transfer decision.
Execute approved work at qualified sites and leave a complete record for inspection, test, and release.
/02 · AI and systems
Choose the lane closest to your work. If your experience crosses roles, select the closest one and explain the range in your application.
Own the cross-layer path from a real customer workflow to private evaluation, agents, software, managed inference, and infrastructure decisions.
Build the product and operating software that turns technical capability into something people can use.
Build the execution layer between a company AI workflow and the infrastructure that runs it.
Turn measured AI-system bottlenecks into correctness-gated GPU kernel, precision, and memory-movement improvements.
Build the verified bridge between host software, programmable logic, memory, and physical devices.
Turn repeated software and infrastructure limits into hardware requirements, experiments, and defensible decisions.
Turn physical limits in compute into materials, process, reliability, and manufacturing research that can be qualified.
Build demand in person: run the events, workshops, field programs, and follow-up that turn technical work into durable relationships.
Learn by shipping a bounded project across agents, software, systems, kernels, hardware, or technical education.
/02
Name the user, the problem, what breaks, and what a useful result changes.
Connect claims to code, traces, experiments, measurements, shipped systems, or reviewed research.
Make the work understandable, testable, and usable by the next person.
/03
Send the clearest evidence of how you work. If uploads fail, email the same material to info@touchdown-labs.com.