STEPHANIE C. JAT

Statistical & Robust Machine Learning

ABOUT ME

I'm a Machine Learning MSc at UCL. Areas of interest: statistically principled learning + evaluation, uncertainty quantification, OOD generalisation, RL as optimal control.

PUBLICATIONS & RESEARCH

ACADEMIC EXPERIENCE

TECHNICAL SKILLS

Python, C, Shell Natural Language Processing Machine Learning ROS, spaCy, hud, MatLab, kooplearn, Linux Data Analysis HTML, CSS, JavaScript, Django English, Cantonese & Putonghua (native) Spanish (Fluent)

EDUCATION

University College London

MSc Machine Learning (2025–2026)

Modules: Supervised Learning, Advanced Topics in ML, Probabilistic and Unsupervised Learning, Approximate Inference, SLT, Reinforcement Learning, Applied DL, Statistical NLP.

University of Cambridge

MPhil Theoretical and Applied Linguistics (2021–2022)

Passed with no corrections [Distinction]. Dissertation in Computational Historical Corpora.

University of Cambridge

BA Theoretical and Applied Linguistics (2018–2021)

Overall 1st with distinction. Dissertation in Forensic Phonetics.

Benenden School, Kent

Secondary School (2013–2018)

GCSE: 13 A*s [English Literature, English Language, Mathematics, Physics, Chemistry, Biology, Latin, Spanish, Chinese, Religious Studies, History, Drama, EPQ]; 1A [Additional Mathematics].

A-level: A*, A*, A, B [English Literature, Latin, Mathematics, Spanish]

PROFESSIONAL & LEADERSHIP EXPERIENCE

Consultant

Tensium, UK (Jul 2026 - present)

  • Build laddered verifiers and adversarial cheat batteries for long-horizon (multi-step, multi-file) RL task environments, targeting shortcut behaviour such as reward hacking, evaluator gaming, memorisation, and specification gaming.
  • Use planted structural mistakes to distinguish genuine competence from surface fixes, evaluating both missed cheats and incorrectly flagged competent behaviour, reflected in a progressive scoring policy.
  • Ship reproducible Python evaluation harnesses with controlled splits, diagnostics, and adversarial stress tests.

Research Assistant

UCL Robotics Lab (Jun 2025 - Sep 2025)

  • Supervised by Dr Valerio Modugno, in collaboration with LAAS-CNRS (Université de Toulouse) to test novel tawaki firmware with their robotpkg software.
  • Installing, repackaging, debugging and writing software to fine-tune the UAV.

AI and Automation Intern

Amber Labs, UK (Oct 2023 - Feb 2024)

  • Built and trained models in Python for internal process automation.
  • Developed a functional chatbot using OpenAI and spaCy.
  • Analyzed data from Monday.com using make.com integrations.

Founder & CEO

Not Just Small Change, UK (Jan 2022 - Present)

  • Successfully launched brand website and two products within three months.
  • Raised over £900 for Eating Disorder Awareness Week and the local Ukrainian crisis appeal.
  • Secured sponsorships and managed sales for fundraising initiatives.
Visit Not Just Small Change →

Editor-in-Chief U-Lingua

The Undergraduate Linguistics Association (Apr 2021 - Apr 2022)

  • Increased reader engagement by 100% through community outreach and social media.
  • Secured £450 grant from the Linguistics Association of Great Britain for hardcopy distribution.
  • Doubled the size of the editorial team, expanding university representation.
Learn more about U-Lingua →

Co-President & Welfare Officer

Pembroke College Committees

  • Organized a sold-out May Ball for 1,300 guests, overseeing all logistics (security, decor, entertainment).
  • As Welfare Officer, increased mental health provisions and launched Unmind/Headspace partnerships.
  • Other notable achievements in fundraising and performance arts.