AI Engineer & Team Lead

I take AI research into production.

I'm a hands-on AI engineering lead who takes research into production at scale. My work spans experimentation, novel data generation, model architecture, reproducible evaluation, MLOps and real-time inference.

See selected work glennonben@gmail.com

Ben Glennon — I lead the researchers and MLOps engineers behind zally's behavioural authentication, taking contrastive-learning models from first experiment to a live platform. Before that, four years at bp building forecasting models and shaping enterprise AI strategy.
**6 yrs**Shipping AI
**zally · bp**Where I've built

Career

Experience

AI leadership, production machine learning, enterprise strategy and applied data science across zally and bp.

zally

Behavioural biometrics. Turning touch and motion patterns into a continuous identity signal.

2025 – present1 yr 4 mos

AI Team Lead
Current2026 – now9 mos
Built and now lead a multidisciplinary AI and Research team spanning model research, MLOps and data quality. Set the technical roadmap and research priorities, support the team's development, and remain hands-on across a behavioural authentication platform supporting 100,000+ concurrent users.
- 100k+ concurrent users
- ~90M signals/sec scored
- Research · MLOps · Data quality
Senior AI Engineer
2025 – 20267 mos
Designed contrastive-learning identity embeddings and established the reproducible research, evaluation and staged-release platform behind zally's Large Behavioural Model.
- Sub-5% median EER
- ONNX on Kubernetes
- Iceberg lakehouse lineage

bp

Global energy. Enterprise AI strategy, responsible AI and applied data science at scale.

2020 – 20254 yrs 9 mos

Technical Program Manager for AI
2024 – 20251 yr
Led enterprise AI strategy and bp's response to the EU AI Act, while building an inventory spanning more than 1,000 AI projects.
- EU AI Act readiness
- 1,000+ project inventory
- Responsible AI by design
Data Scientist
2020 – 20244 yrs
Delivered applied data science across a range of bp teams, from Bayesian production forecasting and subsurface visualisation to founding the AI for Everyone upskilling programme. Graduated with First Class Honours alongside full-time delivery.
- Bayesian production forecasting
- PPFG visualisation
- bp's largest hackathon

Selected technical work

Systems and products I have built.

Examples of my ownership across ML research, infrastructure, mobile data capture and full-stack product development.

01 / 03

zallyBehavioural AI

Behavioural authentication at production scale.

An end-to-end behavioural authentication system: contrastive models, reproducible training data, a researcher-friendly ML platform and low-latency inference on Kubernetes.

PyTorch models exported to ONNX
Kafka-based real-time model harness
Iceberg lakehouse with end-to-end lineage

02 / 03

Behave · Next.js · Phaser

Browser-based behavioural data collection.

Built in a weekend to collect high-quality motion data in the browser, with no app download needed.

Visit behave.game

03 / 03

Bar-Trek · Founder & builder

A full-stack social discovery platform.

A full-stack consumer platform for discovering, creating and sharing pub crawls, growing to 5,000 monthly visitors.

Read the case study

Selected outcomes at bp

Enterprise AI and applied data science.

Before behavioural AI, I worked across modelling, responsible AI, technical strategy and organisation-wide capability building.

1,000+

AI projects mapped

Built bp's enterprise-wide AI inventory, giving global leadership a clear view of maturity and delivery constraints.

EU AI Act

Policy made practical

Led bp's response, translating emerging regulation into standards that engineering teams could actually build against.

First Class

Data science foundation

Delivered Bayesian forecasting and subsurface visualisation while completing a BSc alongside full-time work.

What made all of that possible

The real production dependency behind every model and product above.

3D model by Nishanth · CC BY 4.0

Core capabilities

AI, from research to product.

I work across the full path from experimental model design to data platforms, real-time inference and user-facing products.

That range means fewer hand-offs, faster feedback and systems designed for the real world from day one.

Research → production

Skills matrix.

The tools, methods and engineering disciplines I use to move from an uncertain idea to a dependable product.

ML & deep learning

Model design, representation learning and rigorous evaluation.

Deep learning
PyTorch
Contrastive learning
Bayesian modelling
Hugging Face
Model evaluation

Data platforms

Batch, streaming and governed datasets for research and production.

Spark
Airflow
Kafka
Iceberg
SQL
AWS

MLOps & inference

Reproducible experiments through to low-latency model serving.

Kubernetes
ONNX
MLflow
Real-time inference
Reproducible training

Product engineering

Web, backend and mobile engineering that turns models into products.

Next.js
React
TypeScript
Python
Go
Swift
Kotlin
Phaser

LLM & agentic systems

Building tool-using workflows around reliable retrieval and orchestration.

Tool calling
Multi-agent workflows
LangChain
pgvector
OpenSearch