Jun 2023 – Dec 2024 · 1 yr 6 mos

• + Collaborated across teams to develop and deliver the Control pane agent as a SNAP application – an application containerized with all its dependencies — ensuring smooth and reproducible deployment across Linux distributions.
• + Designed and implemented a one-of-a-kind test infrastructure for NVMe fused operations (fusing together two simpler I/O commands like compare and write) using master-slave architecture and multithreaded programming.
• + Developed and implemented a ring monitoring module enabling the Control Path to periodically track the producer and consumer indices of TX/RX rings and trigger appropriate actions accordingly.

Jun 2022 – May 2023 · 11 mos

• + Designed and implemented a custom I/O latency histogram composed of configurable buckets — each representing a specific latency range (start, end) and tracking the number of I/O operations within that range.
• + Developed a polling mechanism that retrieves I/O statistics from the data plane once every minute to prevent counter rollovers.
• + Ensured fresh and accurate latency metrics by fetching the latest stats at the time of each client request, guaranteeing up-to-date visibility.
• + Exposed the feature through cross-platform RPC APIs, enabling programmatic access to latency stats across components.
• + Built a CLI for easy access and visualization of the latency histogram, improving usability for internal teams and debugging workflows.

Jan 2022 – May 2022 · 4 mos

• + Built an extensive NVMe compliance test suite using the pynvme driver to validate adherence to NVMe protocol 1.4 specifications.
• + Added stress and performance test cases to evaluate reliability and throughput under load.
• + Integrated the test suite into the CI pipeline and configured it to run on daily builds, enabling continuous validation and faster issue detection.

May 2021 – Jul 2021 · 2 mos · Remote · Remote

• + Developed a low-code/no-code database management tool enabling users to perform CRUD operations on SQL and NoSQL databases.
• + Designed a unified, minimal input format that abstracted away database-specific query language.
• + Built a robust multi-layered backend to validate inputs, generate queries, and render results, ensuring data integrity and cross-database compatibility.
• + Implemented a data comparison module that surfaced non-matching records between tables or collections — enhancing debugging and analysis workflows.