Forward Deployed AI Engineer
ThirdAI Automation · San Jose, CA (on-site/hybrid) · [Level: Mid–Senior] · [Salary: $130K-$250K]
About us
When a tool goes down in a fab, everyone panics. A single piece of semiconductor capital equipment can cost tens of millions of dollars and gate an entire production line — and when it starts throwing errors, neither the tool vendor nor the fab knows whose problem it is. Engineers on both sides burn days or weeks chasing correlations through logs and sensor data while wafers pile up and the finger-pointing starts.
That's where ThirdAI comes in. Our Causal Intelligence Platform tells engineers why a tool, process, or defect failed — not just that something changed — cutting root-cause analysis time by 80%+ in production. We work with semiconductor equipment makers, OSATs, and fabs on some of the industry's hardest manufacturing and reliability problems.
The role
You'll own the AI behind live customer deployments — building the models and shipping them into the platform, and working directly with the process and equipment engineers who use them.
Roughly 70% building, 30% in front of customers. That 30% is real. You'll watch engineers investigate defects and diagnose tool problems in their own environment, then turn what you learn into capabilities that work for every customer, not just the one in the room.
We're open on level. Two to three years of exceptional hands-on work can be worth more than ten conventional ones.
What you'll do
- Build and improve causal inference, anomaly detection, and predictive models for root-cause analysis on equipment logs, sensor, SPC, MES, and metrology data
- Own the technical outcome of one or two accounts — from first data pull through validated production results
- Build evaluation frameworks that compare AI-generated root causes against expert judgment and historical outcomes, and measure impact in RCA time, MTTR, and yield
- Write production Python and the data, feature, and inference pipelines behind it
- Make models robust to missing data, noisy signals, and drifting process conditions
- Find the patterns that repeat across customers and turn them into general platform capability
What we're looking for
Must have
- 2+ years building ML or data-intensive systems that reached production, not just notebooks
- Strong Python; comfortable with SQL, APIs, and data infrastructure
- Track record with messy real-world data — you've debugged a pipeline where the data itself was the bug
- You can sit with a domain expert for two hours, understand their problem well enough to disagree with them, and ship something that night
Strong plus
- Causal inference, causal ML, graphical models, or causal discovery
- Time-series, anomaly detection, predictive maintenance, or industrial AI
- Knowledge graphs or graph-based ML
- Semiconductor manufacturing: process control, equipment engineering, yield management, SPC, MES, metrology
- Startup experience with real technical ownership
- Publications, open source, or other evidence of depth
First 90 days
Ship a causal RCA model into a live pilot at a top-tier equipment OEM, validated against the customer's own historical failure investigations, with an evaluation harness the rest of the AI team reuses.
Why this is different
- You'll work on why, not just what — causal reasoning applied to one of the most sophisticated manufacturing environments on earth, with a direct line from prototype to production and no layer between you and the engineers whose problems you're solving.
Apply
www.thirdaiautomation.com