Guiding Your AI Transformation
We don't just build models — we build capability. From standing up your AI department and hiring the right team, to shipping production ML systems and upskilling your engineers in state-of-the-art MLOps, Lighthouse AI takes you the whole way.
What we do
End-to-end AI consulting, not just delivery
We meet you wherever you are — greenfield or scaling — and build the people, processes, and platforms that make AI a lasting advantage.
Build Your AI Department
Stand up an AI/ML function from the ground up — org design, roadmap, tooling, and governance — so AI becomes a durable capability inside your business, not a one-off project.
Talent & Hiring
We source, technically vet, and onboard the right ML engineers, data scientists, and MLOps specialists — building a team that fits your stack, your domain, and your ambitions.
Process & Procedures
Define the standards that make ML repeatable and safe: model lifecycle, experiment tracking, review gates, data governance, and compliance-ready documentation.
MLOps Enablement
We bring your existing team up to speed on state-of-the-art MLOps — CI/CD for models, orchestration, monitoring, and continuous training — through hands-on workshops and pairing.
End-to-End Delivery
From raw data to production inference and automated retraining, we design and ship the full pipeline — scalable, observable, and cost-efficient on your cloud of choice.
Solution Architecture
Serverless and microservices architectures for AI workloads on Google Cloud and AWS — designed for scale, reliability, and predictable cost from day one.
Who we are
Senior AI expertise, backed by a trusted network
Lighthouse AI is a senior-led practice — deep, hands-on engineering, with specialists on call to scale up when a project needs them.
Our founder
Founder & Principal AI/ML Engineer
An engineer by training (MSc), our founder built a career at large engineering companies in Singapore and Germany. They have worked with Python since 2014 — writing research code and numerical solvers for demanding engineering problems: hydrodynamics (wave resistance and vessel response/RAOs), PID control for the dynamic positioning of offshore structures, and Monte Carlo simulations for design operational loads — before moving into AI and machine learning professionally in 2019.
That engineering background shapes how we work: we approach AI the way serious engineering is done — grounded in the data and the numbers, not the hype. Formerly a certified Machine Learning Engineer and Data Engineer.
At a glance
- MSc in Engineering
- 10+ years with Python (since 2014)
- AI & ML professionally since 2019
- Former certified ML Engineer & Data Engineer
- Experience across Singapore, Germany & Bulgaria
- Numerical modelling: FEM, Monte Carlo & PID control
- Also fluent in VBA — incl. a FEM truss solver in Excel
How we scale
Lighthouse AI runs as a lean, senior-led practice backed by a trusted network of AI and data professionals — people we've worked alongside on past projects who are on call to join for larger engagements. You get experienced specialists on demand, matched to your project, without the overhead or hand-offs of a big agency.
Our work
Selected projects
A cross-section of production AI systems we've designed, built, and shipped for clients across finance, industry, marketing, and health.
Creative Optimization Models & MLOps Pipelines
2026 – PresentA US marketing company wanted its ad creatives to perform better without manual guesswork. We are building the models that predict and optimize creative performance — plus the automated pipelines that keep them accurate as new data arrives, so the system keeps improving itself in production.
Surrogate Models & Pareto Optimization for Heap Leaching
2026 – PresentA global mining company wanted to extract more metal from the same ore. We build AI models that predict the outcome of expensive chemistry simulations in seconds instead of hours, then automatically search thousands of operating scenarios to find the settings that maximize recovery — with training and large-scale simulation running on Azure.
Multi-Agent Compliance System for Cross-Border Payments
2025A cross-border payments provider needed to catch compliance risks the moment a transaction happened, not after. We designed and prototyped a system of coordinated AI agents that analyze transactions in real time, and led the client's development team in building it on Google Cloud.
High-Cardinality Document Classification
2025An enterprise needed to classify documents into one of 30,000+ categories — far beyond what a single LLM prompt can handle. We built a multi-stage pipeline (vector search, algorithmic guardrails, and tightly-constrained LLM reasoning) with a human-in-the-loop app, turning an impossible prompt problem into a reliable, self-improving system.
RAG Chatbot for Enterprise Documentation
2024Employees were losing hours searching through internal documentation. We built an AI assistant that answers their questions instantly from the company's own documents — accurate, sourced, and running securely on their own cloud.
Dual-Head Vision Model with Production MLOps
2024A heavy-industry client needed a custom vision model that detects objects, segments scenes, and anonymizes people in live video — running reliably at scale, not just in a notebook. We built the model and the full MLOps pipeline to tune, scale, and visually debug it on Azure Databricks, with training data kept legally clean for commercial use.
Predictive Maintenance for Autonomous Bot Fleets
2021Unplanned failures on structural rails were disrupting warehouse operations. We built a system that detects track defects before they cause grid failures — using dynamic, structure-aware vibration baselines instead of naive global thresholds. The core methodology was patented.
Real-Time IoT System for Smartwatches
2020A wearables company wanted to detect dangerous health events as they happen. We built the end-to-end cloud system — from device data collection to the ML models — that spots falls and anomalies in heart rate and breathing in real time.
FAQ
Questions, answered
A plain-language overview of who we are and how we work — no AI background required.
What does Lighthouse AI do?
We are an AI and MLOps consultancy. We build production machine-learning systems for our clients and help companies build their own AI capability — covering strategy, delivery, hiring, and team enablement.
Can you build an AI department from scratch?
Yes. We help companies stand up an AI/ML function end to end — defining the roadmap and processes, hiring and vetting the right people, and putting the tooling and governance in place so AI becomes a lasting in-house capability.
Which cloud platforms do you work with?
All three major clouds — Google Cloud, AWS, and Microsoft Azure — including their managed ML platforms: Vertex AI, Amazon SageMaker, and Azure ML.
Do you help our existing team adopt MLOps?
Yes. Through hands-on workshops and pairing, we bring your engineers up to speed on modern MLOps — CI/CD for models, orchestration, monitoring, and continuous training — so they can own the systems after we leave.
What kinds of AI problems have you solved?
Across finance, mining, marketing, industrial IoT, and healthcare — from real-time compliance and computer vision to predictive maintenance, document AI assistants, and optimization of industrial processes.
Where are you based and who do you work with?
We are based in Varna, Bulgaria, and work with clients internationally, including in the United States and across Europe.
Get in touch
Let's build your AI advantage
Whether you're starting from scratch or scaling an existing team, we'd love to hear what you're working on. Reach out and we'll get back to you quickly.