I built an LSM-tree key-value storage engine in C++, with a Skip List MemTable, a Write-Ahead Log for crash recovery, and SSTables written to disk. I also added background compaction threads to merge SSTables.
For my predictive telecom traffic routing project, I designed and trained an ST-GNN to forecast traffic spikes 15–60 minutes ahead. I implemented proactive rerouting toward neighbouring towers with available capacity.
In my e-commerce analytics project, I modeled the Olist dataset’s ∼99K orders across nine relational tables in MySQL. My SQL analyses covered revenue growth, cohort retention, RFM segmentation, and product performance; I also built a two-page Power BI dashboard.
At Spark Capital, I cleaned and validated financial datasets for weekly reports, using deduplication, schema validation, and data-type checks. I maintained versioned, documented workflows so reports could be regenerated and audited.

