About Mango AI
AI for Empirical Scientific Computing
Connecting real measurements, domain-specific computation, and verifiable evidence to help engineers understand problems, test hypotheses, and advance investigations.
Measurement-driven scientific and engineering analysis that organizes real observations, specialized computation, and verifiable evidence into traceable workflows.
Mango AI's first product, currently focused on wafer inspection analysis and excursion investigation for semiconductor process and yield engineers.
What Mango AI Focuses On
Many problems in science and engineering require bringing together images, physical measurements, experimental records, and field context, applying the right computational methods, and determining what the existing data actually supports. Statistical scope, units, calculation assumptions, and evidence provenance all directly affect the conclusion.
Mango AI focuses on AI for Empirical Scientific Computing. We organize this process into an executable, traceable, and verifiable workflow: starting from real observations, invoking specialized methods to produce analytical results, explaining what the evidence supports, and explicitly stating what cannot yet be answered.
Here, Empirical emphasizes observation and verification. Statistics, simulation, and approximate computation can all participate while retaining their assumptions, applicable scope, and uncertainties:
Helps understand the engineering question, select appropriate methods and reference materials, assess sufficiency, and separate supported findings from open questions.
Prepares materials, executes deterministic calculations with explicit definitions, records provenance, and validates outputs without relying on language model guesswork.
Inspects raw materials, statistical definitions, and calculation records directly to review findings and make the accountable engineering decision.
Current Product: WaferPilot
We are currently focused on developing our first product, WaferPilot (AI for Wafer Fabrication), starting with wafer inspection analysis and excursion investigation for process and yield engineers.
During routine anomaly investigations, engineers need clear answers to specific questions: What was observed on this wafer? Where are the anomalies concentrated? What statistical definitions were used? Which judgments are supported by current materials, and what evidence is still missing? Around this closed loop, WaferPilot supports four core steps:
Combine registered wafer inspection data, spatial images, and analysis reports to confirm the target object, statistical scope, and applicable methods.
Select appropriate methods for the question and display metrics, spatial distributions, and comparison charts while preserving raw values, definitions, and data sources.
Separate observations, supported conclusions, method limitations, and missing materials; engineers can inspect raw materials directly and ask follow-up questions.
Organize the question, computational results, evidence citations, unconfirmed items, and next-step recommendations into structured material for review and handoff.
Founder & Background
Mango AI was founded by Feng Gao.
Before founding Mango AI, he spent years at Google working on Search and AI projects. Prior to Google, he worked in the semiconductor industry at Synopsys on EDA software development and at Lattice Semiconductor on FPGA design. He holds two Master's degrees—one in Electrical Engineering (EE) and one in Computer Science (CS) with a focus on Machine Learning—and completed the Stanford Graduate School of Business (GSB) LEAD program.
Having worked across FPGA design, EDA engineering computation, and large-scale AI systems, he began exploring opportunities at their intersection in 2023 and transitioned full-time to building the company in 2025. He chose to start with empirical analysis in wafer fabrication because engineering decisions in the fab rely heavily on real measurement data and rigorous computation, where generic AI tools fall short of traceability and verification requirements.
Spent years building and delivering large-scale Search and AI projects.
Worked on EDA software development at Synopsys and FPGA chip design at Lattice Semiconductor.
M.S. in Electrical Engineering (EE) and M.S. in Computer Science (Machine Learning); completed the Stanford GSB LEAD program.
Product & Contact
If you work in wafer fabrication, yield analysis, process engineering, or semiconductor equipment development, we welcome you to explore WaferPilot's current capabilities or reach out by email to discuss specific engineering workflows.