Drew Wolrige
Systems Engineer | Agentic Software & Harness Engineering
I work on complex software and systems problems where architecture, reliability and technical judgement matter.
After close to three decades in technology, my confidence is in getting inside the problem, understanding what is really going on, and working out a practical way forward.
My current focus is agentic software engineering, harness engineering, applied AI systems and trust-first architectures through OpenTendril, AVYN Health and JurnX.
What I Do
I work directly in the product, system or process. I am most useful when something is complex, fragile, unclear, moving quickly, or requires stronger engineering boundaries.
Agentic Systems & Harness Engineering
Engineering the system around probabilistic models: context, tools, permissions, execution boundaries, persistent state, orchestration, verification, observability and human control.
Architecture & System Reviews
Understanding how a system actually works, identifying weak boundaries and structural problems, and determining what needs to change for it to remain reliable, secure and maintainable.
Product Recovery
Getting a derailed, limited, or messy software project back on track. Understand what is broken, what matters, what can be saved, and what needs to change.
Pragmatic Automation & Efficiency
Working out where AI, automation or lightweight software can remove unnecessary work, and where stronger engineering discipline is required.
Current Work
AVYN
## AVYN
AVYN is a privacy-sensitive personal health record designed to help people build and maintain a useful longitudinal health record from the information they already have. It uses AI to turn documents and other source material into structured, reviewable information while preserving the connection to source evidence rather than treating model output as authoritative.
It is also a real-world proving ground for AI-assisted software engineering, where I am testing how probabilistic models and coding agents can contribute materially to a substantial multi-service product without giving up architectural discipline, security boundaries, verification, provenance or human control.
OpenTendril
An applied research and engineering project exploring harness engineering for governed autonomous software work.
OpenTendril engineers the environment around probabilistic models: context, tools, permissions, execution isolation, persistent state, orchestration, verification, observability and human-control boundaries.
The model is an input to the architecture, not the architecture itself.
JurnX
A trust-first records and provenance architecture for systems that need attributable history, reliable evidence and human control.
Inline-Skates
A stake market experiment in niche websites, product discovery, and AI-assisted web builds.
Where I Can Help
Agentic Software & Harness Engineering
Architecture & System Reviews
AI Control, Verification & Reliability
Product Recovery
Distributed Systems & Developer Infrastructure
Pragmatic Automation & Efficiency
Contact
If you are working on a difficult software, systems or agentic-AI problem and need deep practical technical judgement, send me a message.
I am particularly interested in agentic software engineering, harness engineering, AI control and verification, developer infrastructure, distributed systems, trustworthy AI deployment and complex systems problems that do not yet have obvious answers.