Mohamed Atoui
BSc Computer Science and AI, First Class Honours
AI Engineer, London
About
Hi, I'm Mohamed — I was Co-Founder & AI Lead at 
Nordlys Labs, where we built an LLM routing system that scored 75.6% on SWE-bench Verified (December 2025), above the strongest published single-model baseline at the time (Claude Opus 4.5, 74.4%), and scaled to 500+ users.
I'm a Computer Science & AI graduate of 
Royal Holloway, University of London (First-Class Honours, 2026). My research is on transformer attention mechanisms and KV-cache compression for large language models. Alongside that, I've worked on adversarial robustness for ML-based intrusion detection at the ISG Smart Card & IoT Security Centre, sports analytics at 
Stats Perform, taught Object-Oriented Programming & Software Engineering at Royal Holloway, and co-organise the London Systems Club.
More in my CV.
Interests: Machine Learning, LLM Systems, Applied AI, Product Development.
Latest News
Presented MoM Framework at Online Meetup
Presented MoM Framework at Online Meetup
Presented the MoM (Mixture of Models) framework in an online meetup, followed by engaging discussions with the audience on LLM routing and model selection strategies.
Co-organising the London Systems Club
Co-organising the London Systems Club
Started co-organising the London Systems Club, a community bringing together systems-level engineers and practitioners to share expertise, exchange ideas, and learn from one another.
Teaching Assistant at Royal Holloway
Teaching Assistant at Royal Holloway
Started as a Teaching Assistant for Object-Oriented Programming & Software Engineering at Royal Holloway, University of London.
Product Data Intern at Stats Perform
Product Data Intern at Stats Perform
Joined Stats Perform as a Product Data Intern, working on sports analytics and data science.
Founded Nordlys Labs
Founded Nordlys Labs
Co-founded Nordlys Labs, building intelligent LLM routing systems that select the best model for each task. Scored 75.6% on SWE-bench Verified (Dec 2025) against 74.4% for the best single model. Scaled to 500+ users.
Skills
ML & AI
Backend
Frontend
Infrastructure
Languages
Internship & Work Experience
Nordlys Labs
Nordlys Labs
Co-Founder & AI Lead
- Co-founded Nordlys Labs and led ML research on LLM routing — picking the best model per request based on task signals.
- Designed and shipped a routing system that scored 75.6% on SWE-bench Verified (December 2025), above the strongest published single-model baseline at the time (Claude Opus 4.5, 74.4%).
- Scaled the platform to 500+ users; built the production stack on Python / FastAPI / LangChain with Redis + PostgreSQL behind a Dockerised deploy.
- Selected for the Antler accelerator program; presented the MoM (Mixture of Models) framework at an online meetup.
Royal Holloway, University of London
Royal Holloway, University of London
Teaching Assistant
- Teaching Assistant for Object-Oriented Programming and Software Engineering (Year 2).
- Run weekly lab sessions, debug student code in Java, and mark coursework / exam scripts.
- Hold office hours focused on design patterns, testing practice and OO modelling.
London Systems Club
London Systems Club
Co-organiser
- Co-organise the London Systems Club, a community bringing together systems-level engineers and practitioners to share expertise, exchange ideas, and learn from one another.
Thinking About Thinking
Thinking About Thinking
Ambassador
- Ambassador for the Thinking About Thinking community, supporting its events and outreach in London.
Stats Perform
Stats Perform
Product Data Intern
- Analysed large-scale NFL, NBA and MLB datasets in SQL and Python; built feature-engineering and feature-selection pipelines for production analytics products.
- Modelled three-year revenue projections for AWS marketplace distribution using Monte-Carlo simulations, informing go-to-market strategy.
- Worked with enterprise clients to scope analytics deliverables and presented findings to cross-functional product teams.
ISG Smart Card & IoT Security Centre
ISG Smart Card & IoT Security Centre
Machine Learning Researcher
- Research on adversarial robustness of ML-based intrusion detection systems for autonomous vehicles.
- Studied evasion attacks against in-vehicle network IDS models (CAN-bus traffic) and explored defences: adversarial training and input sanitisation.
- Trained PyTorch classifiers on automotive intrusion datasets and benchmarked their behaviour under perturbations.
Academic Services
Object-Oriented Programming & Software Engineering
Get in Touch
Want to chat? Feel free to reach out via email →
- • Ask questions
- • Explore collaboration opportunities