Research

Research

AI-driven operational intelligence for quality and manufacturing systems in deep-tech hardware startups.


Thesis

My thesis investigates how deep-tech hardware startups — companies developing genuinely novel physical products without an established manufacturing history to draw on — can build AI-driven operational intelligence into their quality and manufacturing systems. Legacy industries like automotive or aerospace inherit decades of process capability data, failure libraries, and supplier quality history. Deep-tech hardware startups, by contrast, are often building the failure library in real time, alongside the product itself.

The study uses a mixed-methods sequential explanatory design: a quantitative phase establishes patterns across the industry, followed by a qualitative phase — structured interviews — that explains and contextualizes those patterns from the perspective of practitioners actually running quality operations inside deep-tech hardware companies.

Mixed-methods research Operational intelligence Deep-tech hardware Quality systems Applied AI

Methodology

The design proceeds in two connected phases:

Survey instrument

A structured survey administered to quality and manufacturing practitioners across deep-tech hardware organizations, designed to surface patterns in how AI-driven tooling is — or isn't — being adopted into operational quality workflows.

Interview guide

Semi-structured interviews that explain the quantitative findings in context, drawing out the practical, organizational, and technical constraints that shape how quality intelligence systems actually get built inside fast-moving hardware companies.

Research interests

Beyond the thesis, my broader research interest sits at the boundary between classical reliability engineering and emerging tooling — where methods like Weibull analysis, HALT/HASS, and DOE meet AI-assisted diagnosis, log parsing, and pattern detection in hardware quality data. Two areas I'm following closely:

  • IC design and VLSI — as a longer-term learning direction, deepening the connection between quality engineering and the semiconductor layer underneath the systems I test.
  • Applied scripting for QE workflows — using bash and lightweight tooling for log parsing, hardware diagnostics, and network debugging, in service of faster root-cause work.
Status: Chapter 1 stub sections drafted in full; Chapter 3 outline, interview guide, and survey instrument complete; Chapter 2 citations audited and corrected.