Portfolio

🎓 Education

Universität des Saarlandes, Saarbrücken, Germany Oct 2022 – Oct2025

KJ Somaiya Institute of Technology, Mumbai, India Jul 2018 – May 2022

Max Planck Institute for Informatics, Saarbrücken, Germany Feb 2025 – Oct 2025

Extended the legacy JSVM (H.264/SVC) C++ codebase by implementing bitstream-level optimization modules, enabling fine-grained data reduction without re-encoding.
Designed and implemented a dependency-graph–based ranking algorithm in C++ to selectively remove low-impact data, preserving structural correctness while reducing bitrate by up to 86%.
Built an experimental benchmarking framework to quantify quality–efficiency trade-offs, evaluating optimization strategies across large video datasets under strict decoding constraints.

Max Planck Institute for Informatics, Saarbrücken, Germany Aug 2024 – Mar 2026

Designed and implemented an optical switch simulation model in ns-3 for optical network experimentation.
Reduced IP aliasing analysis runtime by 30% by migrating a pandas-based data processing workflow to Dask and enabling parallel execution across multiple workers.
Developed a Docker-based BGP fuzzing framework to systematically test routing implementations (FRR, BIRD, OpenBGP), automating containerized test orchestration.
Built a multi-process Python pipeline to process RIPE Atlas daily dumps, distributing parsing and processing tasks across worker processes and coordinating stages through inter-process queues for scalable analysis.

OTH Regensburg, Sappz, Regensburg, Germany Jan 2024 – Dec 2024

Implemented query pagination (limit/offset) in the Apache CouchDB Python client, py-couchdb, improving performance when handling large result sets.
Maintained and extended a Python-based ROS testbed management library for orchestrating distributed data collection from heterogeneous sensors.
Engineered a scalable data pre-processing pipeline for regulatory compliance by automating data filtering and analysis, ensuring accuracy for datasets exceeding millions of records.

Implemented a C++ distributed file system with FUSE, featuring a scalable, low-contention client-server model for shared network-backed storage.
Ensured reliability and optimized performance using Valgrind (Memcheck, Callgrind), Address/Thread Sanitizers, and Flame Graph profiling to detect memory leaks, race conditions, and bottlenecks.

Developed a network discovery agent in Rust using LLDP, with frame parsing and generation for protocol compliant device communication and topology mapping.

Developed a probabilistic grammar-based fuzzer in Python for SQLite, generating diverse and semantically valid SQL queries to test database robustness.
Introduced memoization and precomputation of grammar expansion costs, during input generation. Increasing fuzzing throughput by 30%.