Building agentic chat application for Wizzair

We are building an engineer community. Join us!https://growmesh.io

Exceeded answer quality expectations by 15% of a fully customized RAG chatbot for high profile finance firms by leading AI research and implementation utilizing several state of the art techniques. (Langchain, Python, Qdrant, Azure AI Search, OpenAI, NLP natural language processing, DSPy, Neo4j, Arangodb, graph, agents | FastAPI, PostgreSQL, AsyncIO)

- Sped up customer service agent’s domain knowledge search by 60% in most cases by developing a compliant custom RAG chatbot.- Fine-tuned, evaluated, and deployed large language models (LlamaIndex, Python, TypeScript, React, Azure, AWS)

- Boosted advertising revenue by 35-40% by leveraging insights derived from marketing data analysis. Adjusted advertisement targeting and business logic to serve better ads and elevate revenue (statistics, machine learning, decision trees, spatial analytics, natural language processing, keyword analysis, Pandas, Numpy, Scikit-learn)- Freed up on average 8 hours per week for marketing coworkers by developing several in-house real-time dashboards with efficient processing, greatly simplifying decision making; and full-stack web applications that sped up their day-to-day workflows (SQL, Python, Apache Spark, TypeScript, React, Plotly, DigitalOcean, Linux)- Developed a custom full-stack RAG solution used by content writers. The application uses LLMs to generate complete, humanized, and formatted blog posts - including images - based on custom domain knowledge (LangChain, Python, Stable Diffusion, Angular)- Crafted a new data pipeline that amasses up-to-date campaign information from diverse advertising platform APIs. The acquired information facilitates better decision-making concerning advertising costs. (SQL, Python, Linux, DigitalOcean)- Took initiative by innovating company DevOps culture, reducing development time by 10%.

- Accelerated and instituted automated testing for IoT devices in multiple factories by prototyping and engineering full-stack Bluetooth and WLAN software for the tester device (C#, ASP.NET, TypeScript, Angular, Linux, RF communication, A2DP, docker, NI TestStand)- Took on the responsibility of owning the Bluetooth part of the product. Discovered new client needs by mapping customer needs, prototyped new functionality and prioritized feature roadmap.

- Maintained an internal database utilized by over ten automotive companies and fabricated new API endpoints.- Trained convolutional neural network for real-time object detection.- Designed and implemented a new product that uses multiprocessing, networking, cameras, and image processing (MongoDB, Python, ASP.NET, OpenCV, Keras, Microsoft Azure, TypeScript, Linux)

- Completely eliminated (>98%) a critical manual workflows time requirement for 3 colleagues by writing automation software (Office Visual Basic, SQL, Excel). This was the main responsibility of my colleagues, so the teams efficiency skyrocketed as a result. - Sped up a manual Excel workflow for a senior colleague using automation, resulting in a 60% time reduction for this task. This freed up 6 hours per week for this colleague.