Tell me what’s painful in your organization. I’ll build the AI that fixes it.
I’m Alek Lichucki, a senior AI engineer in Los Angeles. By day I help companies roll out AI across the whole organization. I bring the same experience to small and medium-sized businesses: I find the work that is slow, repetitive or easy to get wrong, and I build AI that takes it off your team.
Pains I solve
If one of these sounds like your week, it is fixable.
“My team answers the same questions and repeats the same manual steps every day.”
An assistant that works inside the tools you already use, handles the routine work, and asks before it sends, spends or changes anything.
AI assistants and agents
“Nobody can get a number without waiting on the one person who knows the data.”
A way to ask your business data questions in plain English and get correct answers.
Chat-to-data
“Two teams report two different figures for the same thing.”
A shared record of what your data and metrics mean, so people and AI use the same definitions.
Chat-to-data
“We guess at next month’s sales and which leads to call first.”
Forecasting and scoring models built on your own history, tested honestly, easy to roll back.
Machine learning
“Our data is too sensitive to send to an AI service.”
AI built on open-source models that runs on hardware you own, so nothing leaves the building.
Open-source and local AI
“We pay for AI tools and nobody is using them well.”
A rollout of the tools you already have, such as Gemini Enterprise or Claude, connected to your own data and set up for how your teams work.
Google and Gemini EnterpriseAnthropic and Claude
“We know AI could help and don’t know where to start.”
A short plan in plain language: what to fix, what to build first, and what each step makes possible.
Data strategy
Why me
Enterprise experience, sized for your business
My day job is rolling out AI across companies. You get the same engineer and the same standards without hiring a consultancy.
I have no vendor to sell
I am an expert in Anthropic’s tools, Google’s tools including Gemini Enterprise, and open-source models on your own hardware. I will tell you which one fits your problem and why.
I build things that run
My own AI assistant is in daily use. What I build for you is tested, monitored, and can be undone.
You work with me
I do the work myself from the first conversation to the handover.
Six years, and a master’s from Berkeley
Six years in machine learning and AI, and a master’s in data science from UC Berkeley.
Plain language
I will tell you what AI can and cannot do for you before you spend anything.
How I fix it
I am an expert in three kinds of AI tooling, so the recommendation fits your problem and not a vendor.
Anthropic’s tools
Claude models and the Model Context Protocol, for assistants and agents that work inside your tools.
The private assistant’s tool layer is about 30 tool servers built on the Model Context Protocol.
Google’s tools
Gemini Enterprise, Vertex AI, BigQuery and Google Cloud, for businesses that already run on Google.
My current work is enterprise-wide AI enablement for large companies on Google Cloud.
Open-source models, locally hosted
Open-source models and custom AI on hardware you own, so your data stays in-house and there are no per-message fees.
The private assistant answers every message on hardware I operate myself, on an open-source model, and is in daily use.
Managed tools are faster to start. Open-source models on your own hardware keep your data in-house and remove usage fees. I will tell you which fits.
Selected work
Three substantial systems, self-built. The figures are illustrations, not screenshots.
Forecasting system
Pitch sequencing
Predicting what happens next, live, from millions of past events, and knowing honestly how good the prediction is. The business version is forecasting.
Predicts the next pitch in a live game, measured honestly against simple baselines.
46.5%
top-1 accuracy on the next pitch type for my newest model; the system running live scores 43.3%
2026 regular season, 710,006 pitches
Private AI assistant
Private assistant
Everyday planning was scattered across a dozen apps and one person’s head, and the people using it wanted replies generated on private hardware I operate, not by a cloud AI service.
A private assistant on self-hosted open-source models, in daily use.
About 150
tools across 30 tool servers
Design and method
Knowledge system for chat-to-data
AI writes confident queries that are wrong for a specific business, because it does not know what the company’s data means.
The method for capturing a business’s definitions once and keeping them correct.
Experience
I am an expert in three kinds of AI tooling: Anthropic’s, Google’s (Gemini Enterprise included), and open-source models and custom AI hosted on your own hardware.
2026 to now
Senior AI Engineer
At a cloud and AI consultancy
Enterprise-wide AI enablement for large companies, on Google Cloud and other tools.
2024 to 2026
Data Scientist, Sales and Automation
At a real-estate software company
Sales forecasting and scoring models, chat-to-data products, knowledge bases that ground AI, and tooling for AI agents.
2021 to 2024
Data and Analytics Consultant
At a business and technology consultancy
Marketing analytics for a major streaming service: an A/B testing framework, customer segmentation, churn prevention, anomaly detection and revenue forecasting.
2020 to 2021
Data Science Analyst
At a global IT services firm
Natural-language and deep-learning models, and dashboards for revenue, returns and supply chain.
Education
2024
Master of Information and Data Science
UC Berkeley
B.S. Data Science and Statistics
UC Santa Barbara