Docs change #5
Description
Dr Alex Turner
Senior Principal Software Engineer — Python Runtime, Meta
I solve problems that the consensus says cannot be solved. The method is always the same: find the faulty axiom that everyone else has accepted as given, discard it, and reframe the problem so the solution falls out. Clustered an ultra-low-latency database by abandoning quorum. Replaced Pickle with a secure serialiser by treating it as a declarative problem, not a library swap. Ran millions of CTS tests in hours instead of weeks by parallelising across headset clusters. Four decades, same pattern: if you want convention, don't hire me. If you hate hearing "cannot be done", hire me.
Current Role
Senior Principal Software Engineer in Python Runtime at Meta, focused on:
- AI runtime optimisation for performance
- AI-driven systems-level software development via advanced peer-to-peer collaborative models
Experience
Meta (December 2022 – Present)
Python Runtime — Senior Principal Software Engineer
- Free-threaded Python for AI workloads
- Compiler writing by AI for AI acceleration
- Advanced peer-to-peer collaborative models for AI-driven systems-level software development
Reality Labs — VROS (VR Operating System)
- Automated mass testing of custom hardware — parallelised millions of CTS Android tests across clusters of headsets via mass remote access, reducing test cycles from weeks to hours
- Release policy development and quality standards across Reality Labs
- Automated test creation with AI
Reality Labs — VR Product Health
- Testing and quality
- End-to-end testing process, from Gradle to grave
Bank of America (January 2017 – December 2022)
Senior Principal & Director — Quartz Core
- Mission-critical Python/C++ trading and governance system; multiple systems still in production running billions of transactions per day
- Bespoke share-nothing Sandra database clustering design and implementation — solved a problem considered impossible by abandoning quorum in favour of a leader-follower architecture, preserving ultra-low-latency requirements
- Secure serialisation/deserialisation system as a drop-in Pickle replacement — declarative recursive-descent object-to-linear-and-back pair with high-speed object identity and type-triggered custom deserialisers, audited for security
- Python version-agnostic software frameworks for system upgrades
Morgan Stanley (prior to 2017)
- High-frequency/ultra-low-latency trading system development
- HFT analysis tooling and compiler optimisation
Earlier Career (1997–2010s)
Post-doctoral research at the University of Zurich, followed by two decades spanning pharmaceuticals, dot-com, retail systems, hardware distribution and refresh software, and compiler design (team lead). Earlier career in computer network hardware design (pre-doctorate).
Education
PhD, Computational Quantum Mechanics — University of Bath, 1997
BSc, Chemistry — University of Leicester, 1989
Key Skills
- Finding the faulty axiom: reframing "impossible" problems by identifying and discarding false constraints
- Large-scale and parallel system design for extreme performance requirements
- AI: runtime optimisation, code generation, compiler design by and for AI
- Python and C++ at the runtime/internals level, in mission-critical production systems
- Testing and quality engineering at scale — process design, automation, and AI-assisted test creation