1. 2023 — Present

   AI Knowledge Engineer @ Novo Nordisk

   R&D - Semantic Interoperability · Måløv Capital Region Denmark

   • Technical lead for knowledge-graph initiatives across R&D — ontology design, schema modelling, and graph delivery end-to-end for AI and non-AI use cases.
   • Designs agentic architectures over governed graph data, giving agents reliable, schema-aware access to enterprise knowledge instead of unconstrained retrieval.
   • Leads R&ED Chat, a GenAI assistant for 600+ researchers; introduced Graph RAG over RDF and ontologies, cutting complex-query latency by 82% via semantic subgraph filtering.
   • Tunes hybrid graph + vector retrieval using reranking and Reciprocal Rank Fusion, evaluated on context precision.
   • Built a Protocol Review Service for FAIR compliance: agentic human-in-the-loop workflows that validate metadata and resolve duplicate entities.
   • Delivered a SHACL-validated data-contracts platform replacing manual YAML handoffs, improving pipeline reliability by 45%.
   • Advises engineering and scientific teams on graph adoption; defines usage patterns and governance for shared graph infrastructure.

   Agentic ArchitectureGraph RAGHybrid RetrievalSPARQLSHACLNeo4jMCPPythonLLMs
2. 2020 — 2023

   PhD Researcher @ Aalborg University

   Aalborg, Denmark

   • Researched and engineered scalable graph data transformations and performance-driven query processing, with emphasis on data quality and validation at scale.
   • Published in VLDB, SIGMOD, and EDBT; transferred outcomes into practical data models and reproducible pipelines with engineering and scientific teams.

   Knowledge GraphsQuery OptimizationData QualityJava
3. 2019 — 2020

   Data Science Consultant @ Cassiopeia, Aalborg University

   Aalborg, Denmark

   • Led research and engineering engagements on data-intensive systems, delivering performance improvements through database and pipeline optimization.
   • Produced client-facing technical deliverables — evaluation results, design recommendations, and documentation supporting system redesigns.
4. 2018

   Data Engineer, Intern @ everis NTT Data

   Barcelona, Spain

   Designed scalable data architectures, semantic models, and ETL pipelines for EU-funded research programs.

5. 2016 — 2017

   Software Engineer @ eConceptions

   Islamabad, Pakistan

   Built Java-based backend APIs and telecom value-added-services platforms (SMS, IVR) integrated with operator infrastructure, supporting production mobile apps such as Careem Captains and Jazz Educator.

6. 2015

   Software Engineer, Intern @ eSolPro

   Islamabad, Pakistan

   Built REST APIs in PHP and delivered responsive web modules from design through deployment.

1. Scalable Extraction and Adoption of Shapes for Improving Data Quality and Query Processing in Knowledge Graphs

   PhD Thesis2024Aalborg University · Supervised by Prof. Katja Hose & Prof. Matteo Lissandrini

   Thesis PDF

2. Digital Evolution: Novo Nordisk's Shift to Ontology-Based Data Management

   J. Biomed. Semantics202516 citationsShawn Tan, Shounak Baksi, Thomas Bjerregaard, Preethi Elangovan, Thrishna Gopalakrishnan, Darko Hric, Joffrey Joumaa, Beidi Li, Kashif Rabbani, Santhosh Venkatesan, Joshua Valdez, Saritha Kuriakose

   Publisher

3. Transforming RDF Graphs to Property Graphs using Standardized Schemas

   SIGMOD202524 citationsKashif Rabbani, Matteo Lissandrini, Angela Bonifati, Katja Hose

   Publisher Code Poster

   Abstract

   Knowledge Graphs can be encoded using different data models. They are especially abundant using RDF and recently also as property graphs. While knowledge graphs in RDF adhere to the subject-predicate-object structure, property graphs utilize multi-labeled nodes and edges, featuring properties as key-value pairs. Both models are employed in various contexts, thus applications often require transforming data from one model to another. To enhance the interoperability of the two models, we present a novel technique, S3PG, to convert RDF knowledge graphs into property graphs exploiting two popular standards to express schema constraints, i.e., SHACL for RDF and PG-Schema for property graphs. S3PG is the first approach capable of transforming large knowledge graphs to property graphs while fully preserving information and semantics. We have evaluated S3PG on real-world large-scale graphs, showing that, while existing methods exhibit lossy transformations (causing a loss of up to 70% of query answers), S3PG consistently achieves 100% accuracy. Moreover, when considering evolving graphs, S3PG exhibits fully monotonic behavior and requires only a fraction of the time to incorporate changes compared to existing methods.

4. Mining Validating Shapes for Large Knowledge Graphs via Dynamic Reservoir Sampling

   SEBD20242 citationsMatteo Lissandrini, Kashif Rabbani, Katja Hose

   Proceedings (CEUR Vol-3741) PDF

   Abstract

   Knowledge Graphs (KGs) are databases that model knowledge from heterogeneous domains using the graph data model. Shape constraint languages have been adopted in KGs to ensure their data quality. They encode the equivalent of a schema in the Resource Description Framework (RDF). Unfortunately, few KGs are accompanied by a corresponding set of validating shapes. When validating shapes are missing, the solution is to extract them from the graph via mining techniques. Current shape extraction methods are often incomplete, not scalable, and generate spurious shapes. Thus, in this discussion paper, we present our recent contribution: a novel Quality Shapes Extraction (QSE) method for large graphs. QSE computes confidence and support for shape constraints via a novel Dynamic Reservoir Sampling method, enabling the identification of informative and reliable shapes. QSE is the first method (validated on WikiData and DBpedia) to extract a complete set of shapes from large real-world KGs.

5. SHACTOR: Improving the Quality of Large-Scale Knowledge Graphs with Validating Shapes

   SIGMOD202323 citationsKashif Rabbani, Matteo Lissandrini, Katja Hose

   Publisher PDF Website Demo video

   Abstract

   We demonstrate SHACTOR, a system for extracting and analyzing validating shapes from very large Knowledge Graphs (KGs). Shapes represent a specific form of data patterns, akin to schemas for entities. Standard shape extraction approaches are likely to produce thousands of shapes, and some of those represent spurious constraints extracted due to the presence of erroneous data in the KG. Given a KG having tens of millions of triples and thousands of classes, SHACTOR parses the KG using our efficient and scalable shapes extraction algorithm and outputs SHACL shapes constraints. The extracted shapes are further annotated with statistical information regarding their support in the graph, which allows to identify both erroneous and missing triples in the KG. Hence, SHACTOR can be used to extract, analyze, and clean shape constraints from very large KGs. Furthermore, it enables the user to also find and correct errors by automatically generating SPARQL queries over the graph to retrieve nodes and facts that are the source of the spurious shapes and to intervene by amending the data.

6. Extraction of Validating Shapes from very large Knowledge Graphs

   VLDB202388 citationsKashif Rabbani, Matteo Lissandrini, Katja Hose

   IFIP TC2 Manfred Paul Award 2024

   Publisher Extended version Website

   Abstract

   Knowledge Graphs (KGs) represent heterogeneous domain knowledge on the Web and within organizations. There exist shapes constraint languages to define validating shapes to ensure the quality of the data in KGs. Existing techniques to extract validating shapes often fail to extract complete shapes, are not scalable, and are prone to produce spurious shapes. To address these shortcomings, we propose the Quality Shapes Extraction (QSE) approach to extract validating shapes in very large graphs, for which we devise both an exact and an approximate solution. QSE provides information about the reliability of shape constraints by computing their confidence and support within a KG and in doing so allows to identify shapes that are most informative and less likely to be affected by incomplete or incorrect data. To the best of our knowledge, QSE is the first approach to extract a complete set of validating shapes from WikiData. Moreover, QSE provides a 12x reduction in extraction time compared to existing approaches, while managing to filter out up to 93% of the invalid and spurious shapes, resulting in a reduction of up to 2 orders of magnitude in the number of constraints presented to the user, e.g., from 11,916 to 809 on DBpedia.

7. End-to-End Incremental Data Integration via Knowledge Graphs

   SWJ20249 citationsJavier Flores, Kashif Rabbani, Sergi Nadal, Cristina Gómez, Oscar Romero, Emmanuel Jamin, Stamatia Dasiopoulou

   SWJ PDF

   Abstract

   Data integration, the task of providing a unified view over a set of data sources, is undoubtedly a major challenge for the knowledge graph community. Indeed, such flexible data structure allows to model the characteristics of source schemata, rich semantics for the global schema and the mappings between them. Yet, the design of such data integration systems still entails a manually arduous task. This becomes aggravated when dealing with heterogeneous and evolving data sources. To overcome these issues, we propose a fully-fledged semi-automatic and incremental data integration approach. By considering all tasks that compose the end-to-end data integration workflow (i.e., bootstrapping, schema matching, schema integration and generation of querying constructs), we are able to address them in a unified manner. We provide algorithms for each task, as well as theoretically prove the correctness of our approach and experimentally show its practical applicability.

8. SHACL and ShEx in the Wild: A Community Survey on Validating Shapes Generation and Adoption

   WWW202258 citationsKashif Rabbani, Matteo Lissandrini, Katja Hose

   Publisher PDF Website

   Abstract

   Knowledge Graphs (KGs) are widely used to represent heterogeneous domain knowledge on the Web and within organizations. Various methods exist to manage KGs and ensure the quality of their data. Among these, the Shapes Constraint Language (SHACL) and the Shapes Expression Language (ShEx) are the two state-of-the-art languages to define validating shapes for KGs. Since the usage of these constraint languages has recently increased, new needs arose. One such need is to enable the efficient generation of these shapes. Yet, since these languages are relatively new, we witness a lack of understanding of how they are effectively employed for existing KGs. Therefore, in this work, we answer How validating shapes are being generated and adopted? Our contribution is threefold. First, we conducted a community survey to analyze the needs of users (both from industry and academia) generating validating shapes. Then, we cross-referenced our results with an extensive survey of the existing tools and their features. Finally, we investigated how existing automatic shape extraction approaches work in practice on real, large KGs. Our analysis shows the need for developing semi-automatic methods that can help users generate shapes from large KGs.

9. Optimizing SPARQL Queries using Shape Statistics

   EDBT202125 citationsKashif Rabbani, Matteo Lissandrini, Katja Hose

   Publisher PDF Website

   Abstract

   With the growing popularity of storing data in native RDF, we witness more and more diverse use cases with complex SPARQL queries. As a consequence, query optimization — and in particular cardinality estimation and join ordering — becomes even more crucial. Classical methods exploit global statistics covering the entire RDF graph as a whole, which naturally fails to correctly capture correlations that are very common in RDF datasets, which then leads to erroneous cardinality estimations and suboptimal query execution plans. The alternative of trying to capture correlations in a fine-granular manner, on the other hand, results in very costly preprocessing steps to create these statistics. Hence, in this paper we propose shapes statistics, which extend the recent SHACL standard with statistic information to capture the correlation between classes and properties. Our extensive experiments on synthetic and real data show that shapes statistics can be generated and managed with only little overhead without disadvantages in query runtime while leading to noticeable improvements in cardinality estimation.

10. ODIN: A Dataspace Management System

    ISWC201920 citationsSergi Nadal, Kashif Rabbani, Oscar Romero, Shumet Tadesse

    Publisher PDF Website

    Abstract

    ODIN (On-demand Data Integration) is a system that supports the incremental pay-as-you-go integration of data sources into dataspaces and provides user-friendly querying mechanisms of the resulting dataspaces. ODIN's novelty lies in a largely automated bottom-up approach (i.e., driven by the sources at hand) that includes the user in the loop for disambiguation purposes. ODIN relies on the concept of traceability graphs, which are generic metadata abstractions (i.e., not tailored for a specific task) about the integration of a particular set of data sources. From these graphs, ODIN is capable of generating target-oriented metadata constructs. In this demonstration we focus on those for query answering over dataspaces.

11. ARDI: Automatic Generation of RDFS Models from Heterogeneous Data Sources

    EDOC20196 citationsShumet Tadesse, Cristina Gómez, Oscar Romero, Katja Hose, Kashif Rabbani

    Publisher PDF

    Abstract

    The current wealth of information, typically known as Big Data, generates a large amount of available data for organisations. Data Integration provides foundations to query disparate data sources as if they were integrated into a single source. However, current data integration tools are far from being useful for most organisations due to the heterogeneous nature of data sources, which represents a challenge for current frameworks. To enable data integration of highly heterogeneous and disparate data sources, this paper proposes a method to extract the schema from semi-structured (such as JSON and XML) and structured (such as relational) data sources, and generate an equivalent RDFS representation. The output of our method complements current frameworks and reduces the manual workload required to represent the input data sources in terms of the integration canonical data model. Our approach consists of production rules at the meta-model level that guarantee the correctness of the model translations. Finally, a tool for implementing our approach has been developed.

Master's thesis: Supporting the Semi-Automatic Creation of the Target Schema in Data Integration Systems (BDMA, 2019) — PDF. Full list on Google Scholar.