Hey there 👋

I'm Marian Lambert

AI & Software Consultant

I enjoy building software solutions that create real value for people and businesses, in particular by leveraging the power of AI and LLMs. I am also a guest lecturer in the field of AI and Data Science.

Let's connect on LinkedIn!

Current projects

Undisclosed Project #3

This project focused on building a proof of concept (PoC) for an AI-powered system to improve the efficiency of creating medical assessments. The solution integrated transcription, generative AI, and user-friendly tools, all under tight time and resource constraints.

Negami

Negami is a mobile application that uses augmented reality and gamification to treat patients suffering from neglect after a stroke.

See all projects

Recent Talks

Me, Myself & AI

e-fellows.net Career Evening IT | Online | Oct 17, 2024

The keynote, targeted at young professionals in the field of AI, focused on the latest advancements in the field of artificial intelligence, as well as achievements, current challenges, and future developments. It also provided advice on what individuals could do to thrive in a world shaped by AI.

e-fellows Programm LinkedIn Post Slides

Build Your own LLMs: Customization and Hosting of AI Models

IT After Work | Pforzheim, Germany | Dec 6, 2023

In this talk, I discussed the customization and hosting of Large Language Models (LLMs) in the business context. The presentation highlighted strategies for personalization of LLMs, including retrieval-augmented generation and fine-tuning.

LinkedIn Post XPACE Blog Post Slides (German)

Evaluating Large Language Models and Prompt Variants on the Task of Detecting Cease and Desist Violations in German Online Product Descriptions

DataMod 2023 | Eindhoven, Netherlands | Nov 6, 2023

As part of the KIVEDU project, this talk demonstrated the results of our research on the evaluation of Large Language Models (LLMs) and prompt variants on the task of detecting cease and desist violations.

LinkedIn Post Program

Robustness Analysis of Machine Learning Models using Domain-Specific Test Data Perturbation

EPIA 2023 | Azores, Portugal | Sep 8, 2023

In this talk, I presented our research on robustness testing machine learning models in the image, audio and text domain. The presentation highlighted the importance of domain-specific test data perturbation for the evaluation of robustness.

LinkedIn Post Program

See all talks

Latest Blog Posts

Deploying a Complex Data Pipeline on Azure: Dagster vs. Prefect

Exploring the best options for deploying a complex data pipeline on Azure, comparing Dagster and Prefect.

Azure

Data Pipeline

Dagster

Prefect

Typescript Tricks: Branded Types

Typescript

Typescript Tricks: Defining Types from Arrays

Typescript

See all posts

A High-Level Guide to GPU Utilization (by Charles Frye)

A guide to maximizing GPU output by understanding key utilization metrics—GPU Allocation, Kernel, and Model FLOP/s—and implementing practical improvement steps.

GPU

performance

machine learning

inference

cloud computing

Doordash and Pizza Arbitrage (by Ranjan Roy)

An entertaining yet somewhat sad analysis of the economics of delivery services, highlighting why they can be problematic business models.

business

economics

startups

food delivery

Post-Chat UI (by Allen Pike)

A collection of emerging UI patterns beyond traditional chat interfaces that leverage LLMs, offering inspiration for building new products.

LLMs

product design

See all links