Valerie's Portfolio
AI Engineer at IBM
I turn exploratory ideas into production-bound AI for clients across financial services, telecoms and hospitality. This is where I take the foundations apart, from agentic pipelines to the algorithms underneath, and show how they actually work.
About Me
Hi, I'm Valerie 👋
I'm an AI Engineer at IBM with over two years building agentic and foundation-model systems for clients across financial services, telecoms and hospitality, and I'm in the second year of an MSc in AI and Data Analytics after an earlier MSc in Artificial Intelligence at King's College London. What drives me is seeing a business truly thrive. Not just hit a metric, but grow in a way that lasts. That's why I do what I do: I turn data into decisions that move the needle, helping the businesses and industries that provide jobs and power the economy. Data never lies; it just mumbles sometimes, and AI is helping me uncover those truths faster, so the businesses I work with can act on them sooner. This portfolio is a glimpse of what excites me: the questions I chase, the projects I've taken on, and what's possible when data is put to work. More than anything, I hope it inspires fresh questions about what data is quietly trying to tell us.
Experience
AI Engineer at IBM with 2+ years specialising in foundation models, agentic AI and enterprise AI solutions.
AI Engineer · IBM
Nov 2023 - present- Design, build and deliver AI-powered MVPs and proofs-of-concept on foundation models (LLMs) for clients across financial services, telecommunications and hospitality, turning exploratory ideas into adopted, production-bound solutions.
- Build agentic workflows: LLM agents with tool-calling, orchestration and RAG over enterprise data, to automate complex, multi-step business processes.
- Clean and manage large datasets, integrate models with enterprise platforms (CRMs, data lakes) and run evaluation and performance testing for accuracy, robustness and scalability.
- Founded and lead IBM's internal AI Innovation community, a cross-organisation forum for research discussion, and regularly present AI and data fundamentals to drive technical upskilling.
Education
Level 7 AI & Data Specialist apprenticeship, second year.
Pattern Recognition, Machine Learning, Computer Vision, AI Planning, Multi-Agent Systems.
The commercial lens I still bring to every model I build.
Skills
Featured Projects
Explore my portfolio of AI and data science projects, from machine learning models to data visualizations
Project Example: An Agentic Front Desk — LangGraph + MCP
An inbound request becomes a booked, emailed, calendared appointment — handled by an agent that reasons about each call and takes the actions itself through MCP tools.
Project Example: Predictive Maintenance on an Oil Rig with IBM Maximo
Aging rig machinery, sensed and risk-scored, turned into prioritized work orders: preventive on a schedule, predictive on a threshold, corrective on failure.
Fine-Tuning with InstructLab
How I teach a base model new things from a handful of examples: contribute seeds, let it generate thousands, train, and measure the lift.
Handwriting Recognition — TensorFlow CNN
A convolutional net reading handwritten digits — one of my projects for Pattern Recognition, Neural Networks & Deep Learning at King’s.
Backpropagation, Step by Step
The algorithm I had to memorise, made hands-on: push an input through, watch the error flow back, nudge the weights.
Neural Net Learning a Decision Boundary
A little neural net, hand-written in NumPy, learning to separate two spirals.
Gradient Descent — Learning Rate Race
Three learning rates dropped on the same loss surface — one crawls, one glides, one overshoots.
AI Temporal Planning, the Lasagne Problem
The lasagne question I drilled before my King's AI Planning exam: overlap every cooking step and the same meal is done in 95 minutes instead of 137.
A* Pathfinding
A* finding its way out of a random maze. You can watch the frontier reach toward the exit.
Breadth-First Search: A Cat Finds the Fish
A cat in a hedge garden finds the fish. BFS fans out in rings until it bumps into the goal, then lights up the shortest way there.
Genetic Algorithm — Traveling Salesman
The travelling salesman problem, solved by letting a bunch of routes evolve.
Boids — Emergent Flocking
Three simple rules per bird, and a flock shows up on its own — no leader.
Conway's Game of Life
Four trivial rules per cell, and somehow you get gliders, guns and oscillators.
L-System: A Flower Grown From One Rule
A single rewriting rule, applied to itself a handful of times, grows into a branching flowering plant. The whole bloom is recursion made visible.
Sorting Garden
A messy bed of flowers, each a different height, tidied into a neat row by quicksort. Watch the meadow sort itself shortest to tallest.
Queue and Stack: Cats in Line, Cats in a Pile
Five cats go in, five cats come out. A queue serves them in the order they arrived; a stack serves them backwards. Same cats in, opposite order out.
Get In Touch
Interested in collaboration or have a question? Feel free to reach out!
Let's Connect
I'm always open to discussing new projects, creative ideas, or opportunities to be part of your vision.
Quick Response
I typically respond within 24-48 hours. For urgent inquiries, please reach out via email directly.
Looking for my CV? Just drop me an email and I'll send it over.