Jakub Poźniak

AI / GenAI Engineer

[email protected] · khal-dev.com · linkedin/khalsci · github/KhalSCI · kaggle/jakubkhalponiak

Summary

AI/GenAI Engineer specializing in LLM-based agents, Retrieval-Augmented Generation (RAG) and agentic engineering for enterprise applications. Experienced in natural-language interfaces over structured data (DAX/SQL), Azure AI Search retrieval pipelines and production GenAI deployments on Azure. Grounded in data science, with a track record of shipping products end-to-end, including a cross-platform AI mobile app.

Technical skills

GenAI & ML
LLM agent design, agentic engineering, RAG pipelines, prompt engineering, multi-agent systems, vector search, data science
AI / ML frameworks
Semantic Kernel, LangChain, LlamaIndex, OpenAI API, OpenRouter, Hugging Face, scikit-learn, TensorFlow
Programming & web
Python, C/C++, SQL, DAX, JavaScript, HTML/CSS, React, Redux, Flask, Django
Azure & cloud
Azure AI Search, Azure DevOps, Azure Web Apps, Azure Portal, AWS S3, GCP
Data libraries
pandas, NumPy, Matplotlib, GeoPandas, osmnx, Folium
AI-assisted dev & tools
Claude Code, Cursor, GitHub Copilot, Git, Docker, Jupyter, VS Code, CI/CD
Languages
Polish (native), English (C1)

Experience

Python GenAI Developer

Apr 2025 - present

Accenture Poland · remote

  • Designed and built LLM agents that retrieve structured data through DAX and SQL queries, delivering natural-language interfaces over Power BI datasets and relational databases.
  • Built and tuned RAG pipelines on Azure AI Search, creating custom indexes, schemas and skillsets across diverse enterprise use cases, plus automated tooling to generate data-model schemas so agents interpret complex relationships accurately.
  • Shipped reusable agent plugins and Azure Web Apps to production, integrated into enterprise workflows via Azure DevOps CI/CD pipelines.

python · azure ai search · azure devops · azure web apps · semantic kernel · openai api · sql · dax

Data Scientist

Dec 2024 - Mar 2025

Accenture Poland · remote

  • Performed geospatial analysis for an FMCG client, generating POIs and interactive maps to guide market expansion; built ML models to rank SKUs and points of sale.
  • Designed planograms and assortment recommendations to optimize retail shelf space; created predictive models to forecast business-unit KPIs.

python · sql · geopandas · folium · scikit-learn · pandas · numpy · osmnx

Projects

Rybol: AI fishing journal

2025 - present

computer vision · LLMs · geospatial · mobile

  • Founder and developer of a cross-platform (iOS/Android) AI fishing journal: users photograph a catch and computer-vision AI auto-fills species with estimated weight and length, while GPS, location and weather are captured automatically.
  • Built LLM-generated "what's biting" briefings over a map of fishing spots using OpenStreetMap geodata, plus an atlas of 328 species, season statistics, offline-first sync and a freemium subscription model.

GSMArena scraper & smartphone dataset

2024

scrapy · pandas · seaborn

  • Built a Scrapy scraper (rotating proxies, randomized headers) and published a cleaned dataset on Kaggle (35+ reviews, 10/10 usability score, 10+ community notebooks), then analyzed market and pricing trends with pandas and Seaborn.

Education

Politechnika Poznańska

Oct 2023 - Feb 2027

B.Eng. in Computer Science · Poznań, Poland

Szczecińskie Collegium Informatyczne

Sep 2019 - Jun 2023

IT Technician · Szczecin, Poland

Interests

Semi-professional gaming: Poland's 2nd-division League of Legends and collegiate esports (communication & teamwork under pressure).