Senior / Staff AI Model Engineer
Skills
Computer ScienceProgramming LanguagesBank ConfirmationComplex SystemsCuratingLabelsLarge Language Models
What the job involves
The main requirements, responsibilities and hiring steps.
Requirements
- At least 8 years of experience shipping production software
- Strong proficiency with any programming language
- Strong knowledge of computer science fundamentals testing methodology and systems design
- Experience building evaluation frameworks test harnesses and benchmark suites for complex systems
- Experience running model improvement cycles including dataset curation labeling QA offline experimentation and deploying changes with measurable impact
- Ability to define metrics build measurement pipelines and drive engineering and product decisions from data
- Comfort working across the stack debugging model and tooling failures instrumenting services and partnering with frontend and product on UX patterns that improve safety and trust
- High degree of self motivation and willingness to jump into unfamiliar areas to solve problems
Nice to have
- Fine tuning experience
- Preference optimization experience
- Distillation experience
- Prompt reliability techniques
- Trading domain knowledge
- Rust performance focus
- ML incident response experience
Day to day
- Own the reliability and quality bar for an AI copilot embedded in a trading platform used by sophisticated investors.
- Design and build evaluation systems that measure correctness safety latency and regression risk across market analysis portfolio and risk reasoning and trading workflows including order placement.
- Develop benchmarks including golden sets scenario suites stress and adversarial cases and continuously refreshed market and regime based test corpora.
- Build automated quality gates regression workflows monitoring alerting and incident response processes that protect release quality and support fast fix-forward recovery.
- Partner with engineering and product to define safe tool and action contracts improve model behavior and advance trading domain accuracy and user trust.