Nikola Tanković

I built an advanced conversational AI assistant to deliver seamless, real-time interactions using text and voice. Leveraging the latest AI models from AWS Bedrock and Azure OpenAI, this solution provides context-aware responses by integrating multiple knowledge sources, from hundreds of diverse documents (PDF, PPT, DOCX, ...) to relational databases.

On the technical side, the system features a React front end and FastAPI back end, with bidirectional text and voice streaming. The project successfully applied advanced Retrieval-Augmented Generation (RAG) and graph-based RAG techniques.

The Skillscape project was done for the Scalable Cooperation group at the MIT Media Lab. This project is part of a broader effort to help society understand how artificial intelligence and robotic automation can impact human labor.

My role was to extract and merge data from heterogeneous datasets and visualize them within an interactive web application using multiple D3.js techniques.

A project of science and law where network science and machine learning help to protect vulnerable minorities in a rapidly changing world.

Constitutionology is a neologism for the scientific analysis of national constitutions. Techniques such as information retrieval, network analysis, and natural language processing allow us to understand how constitutions change and what causes specific provisions, e.g., protection from child labor or the right to privacy, to be adopted.

Cash Register is a multi-platform software-as-a-service (Saas) application for issuing invoices on the go that also supports Windows, Android, and iOS platforms.

Technologies: Java/PostgreSQL for the back-end, JavaScript for the web-based front-end, and Java/Android for the native part of the hybrid mobile application.

Edmond is a revenue management optimization tool used in the hospitality industry.

My role was the implementation of the microservices for forecasting occupancy and proposing optimal rates.

Evolving complex software systems (EVOSOFT) have become a central part of a rapidly growing range of applications, products, and services supporting daily human activities from all economic sectors.

As they are often distributed, heterogeneous, decentralized, and inter-dependent and operating in dynamic and unpredictable environments, availability and reliability become key properties for its operation and future evolution.

The novel-and-still-unexplored area of research addressed in this project aimed to understand how abstract software structures and local system properties influence fault distributions, thus affecting mission-critical system properties (among which concerned availability, reliability, and the development of innovative approaches for smart management of their operation and evolution).

My Role: Researcher

This thesis introduces ElaClo, a framework for optimizing application configuration in a cloud environment. ElaClo optimizes the software assembly phase from automatically extracted application models. ElaClo was implemented as an expert tool and evaluated on a real-life cloud application from the retailing business domain where ElaClo was applied to select optimal configurations with regards to service response time objectives and infrastructure costs.

Contributions of this thesis are a model of the execution environment for optimizing cloud information system with regards to quality and infrastructure cost, a component-based information system architecture for dynamic reconfiguration, a method for evaluating the performance and cost and selecting configurations compliant to service level agreement, and evolution algorithm for optimizing configurations in the simulation environment.

I developed Android and iOS mobile applications in the Angular/Ionic 3 framework. It connects to a Python-based back-end that enables live trading—aggregating the current prices, offers, and orders across several cryptocurrency exchanges.

Apache Lucene is a free-and-open-source information retrieval software library, originally written completely in Java by Doug Cutting. It is supported by the Apache Software Foundation and is released under the Apache Software License. I worked as an open-source contributor for Apache Lucene during the Google Summer of Code 2011 and 2012 program.