Piotr Laczkowski

I have developed many python projects in order to automatize various tasks (data-analysis, reporting, plotting etc..), to control sophisticated measurement pipeline with a simple used interface (GUI) or centralize the acquired data in a dedicated database.

Build a reverse image search engine to be able to visually recognise a given product based purely on its image.

This project was related to a smart pricing engine for a pseudo-unique items sold on an e-commerce platform. This project made use of various models ranging from more to less explaiable: i.e. Probabilistic Bayesian Regression..

The goal of this project was to match a text provided by the client to a product on the e-commerce platform. The main difficulty was related to the fact that this text was a free entry field and necessary information needed to be extracted directly from it. Support for 6 different languages and various type of products needed to be implemented.

Developed an advanced time series forecasting system for inventory management, reducing stockouts by 30% and improving supply chain efficiency.

Project aim was to extract colors (and their precise vectors RGB) of an object in the provided photo.

Project aim was to separate the object from its original background and represent it on the unified white background following the style-guide of the e-commerce platform.

Project aim was to detect a potential payment fraud on the e-commerce platform.

Project aim was to rank the most important products to be curated on the backend of an e-commerce platform and focus the efforts needed to account for potential problems in a smart way (increasing the total GMV).

Physical store recommendation engine with advanced ranking, allowing to curate the catalogue and reduce the stockouts by 30%. The system was able to recommend the most suitable products to a customer based on the previous purchases and the current products available.

Developped advances data preprocessing library (OPEN SOURCE) based on Keras preprocessing layers that can be used as a standalone model or incorporated to Keras model as first layers.

Project aim was to develop an advanced non-supervised anomaly detection system based on AutoEncoders. The system was able to detect anomalies based on historical data statistics and included all the pre and post processing withinthe model.

I have developed many LLM based covenrsational systems with several interesting architectures: RAG, Graph-RAG, etc. using various technoligies

I was supervising development of a chatbot that would be able to act as a virtual negociation coach for our staff.

This project aim was to develop collaborative agents that will be able to interact to achieve overall goal while focusing on specific tasks.

Project goal was to be able to detect some non standard patterns in the codebase and alert the team in order to improve the quality of the code and prevent secrets, private data etc. leaks.