About Me
Hi, I'm Jakub Lewandowski, a software engineer specializing in AI-powered applications.
I began my career as a fullstack developer, focusing mainly on backend systems built with NestJS. As the products I worked on increasingly relied on AI, my work shifted towards integrating it, which led me to the Python ecosystem, where most of today's AI tooling lives. Now I work across the whole stack of GenAI applications: AI agents, real-time voice assistants and LLM-based document processing, together with the APIs and cloud infrastructure that run them in production.
I currently work as an AI Specialist at Allianz as part of the Allianz Central Europe Graduate Programme, and I'm pursuing a Master's degree in Computer Science at Warsaw University of Technology.
Outside of work, I study machine learning from first principles, building and training models from scratch to understand how they work under the hood. In my free time, I enjoy playing badminton, and I'm into football and American rap.
Skills
AI & Data
Python
LangGraph
LangChain
Gemini
Vertex AI
Pipecat
Deepgram
Claude Code
Pandas
Backend
Go
FastAPI
NestJS
Node.js
Express
Prisma
PostgreSQL
MongoDB
Frontend
TypeScript
React
Next.js
Tailwind CSS
Material UI
Framer Motion
Cloud, DevOps & Testing
GCP
AWS
Docker
Terraform
Ansible
CircleCI
Jest
Vitest
Cypress
Career Timeline
AI Specialist @ Allianz
Selected for the Allianz Central Europe Graduate Programme from more than 1,500 applicants following a rigorous, multi-stage recruitment process. The 16-month programme combines hands-on work across teams and rotations with structured training and mentoring. I currently focus on voice-based AI agents.
Fullstack GenAI Developer @ Capgemini Invent
Delivered GenAI solutions for clients, combining fullstack engineering with AI integration across two projects:
- Health & safety platform: designed and built the backend API, data model and Google Cloud infrastructure from the ground up, and integrated AI capabilities including data extraction from handwritten inspection reports and incident severity assessment.
- Real-time voice AI agent: developed the administration panel for managing and testing the agent's tools, contributed to the agent's development and deployed it on Google Cloud. The agent was built on real-time models (Gemini Live) to enable natural, low-latency conversations.
Python Developer @ BlockWise
Contributed to a weather forecasting system that combines multiple weather models through ensembling to improve forecast accuracy.
Fullstack Developer @ Mood Up
Developed and maintained web applications using React and TypeScript for frontend implementation. Built and integrated APIs using NestJS framework. Wrote comprehensive unit tests for both frontend and backend layers using Vitest and Jest. Implemented end-to-end testing with Cypress for full application workflow validation. Created CI/CD pipelines and managed production deployments. Planned and designed infrastructure architecture. Integrated backend systems with Large Language Models (LLMs) for enhanced application functionality.
Tutor @ AMBITNI
Teaching programming to students and preparing them for computer science final exams. Focusing on developing practical coding skills and theoretical knowledge.
Health & Safety Incident Platform
Backend and cloud infrastructure for an occupational health and safety platform, featuring incident reporting, AI-assisted document extraction and incident severity assessment. Designed and delivered from the ground up, from the data model to deployment on Google Cloud.
Real-Time Voice AI Agent
A conversational, voice-driven AI agent supporting in-store customer service. It responds in real time and performs actions through tool calls, such as checking product availability or order status. I developed its administration panel, contributed to the agent itself and deployed it on Google Cloud.
Plutus CLI
Plutus is a fast, local-first CLI tool written in Go designed to replace manual spreadsheets for tracking long-term investments. It automates portfolio valuation by concurrently fetching market data and securely storing it in a local SQLite database for easy terminal-based management.
Leftovers
Leftovers is a comprehensive fullstack recipe management platform developed from scratch during my internship at Moodup. It integrates a modern frontend (React, TypeScript) with a robust backend (NestJS, PostgreSQL, AWS S3) and is backed by thorough testing and automated CI/CD pipelines (Docker, CircleCI) to ensure system reliability.