Tech Stack: C#, Python, PostgreSQL, Azure Functions, REST APIs, Docker

• Automated transactional data synchronization between campus point-of-sale and payroll systems using C# Azure Functions, eliminating a 10-hour/week manual CSV reconciliation process prone to human entry errors

• Isolated thread contention during a 5,000-user enrollment spike via jstack; tuned pool logic to eliminate 504 errors that previously blocked ~15% of students

• Established a SemVer-based versioning strategy for internal REST APIs serving 4 campus departments, eliminating 'dependency hell' where uncoordinated schema changes previously stalled weekly deployments

• Identified a nested loop join on an unindexed 2M-row audit table that was timing out reporting dashboards; implemented partial GIN indexes to support JSONB search, reducing p95 latency by 30%

• Caught silent data inconsistencies caused by missing transaction boundaries in Azure Function executions, adding explicit commit and rollback handling to eliminate stale-read failures across enrollment cycles

Tech Stack: Go, C++, PostgreSQL, gRPC, REST, Docker, Linux

• Engineered Go and C++ telemetry ingestion services for industrial PLCs; maintained consistent 600 RPS throughput during burst device-sync windows across distributed nodes

• Investigated burst-load latency incidents using pprof, trace data, and runtime profiling, isolating CPU hot paths and request-path bottlenecks that reduced p99 latency from ~400 ms to ~80 ms

• Profiled PostgreSQL query plans on a 10M-row time-series table where missing composite indexes caused dashboard timeouts; redesigned schema to use partitioned tables, cutting p95 read latency by 35%

• Caught a silent API contract violation during integration review, where an undocumented field removal in a gRPC response would have broken downstream compatibility before release

• Traced a serialization bottleneck in gRPC response encoding under burst traffic, replacing reflection-based marshaling with pre-generated protobuf code and cutting per-request allocation overhead by ~40%

Tech Stack: Java, Python, Kafka, PostgreSQL, Linux

• Improved consumer throughput by ~40% by adding partition-aware routing (consistent hashing) after noticing uneven partition utilization and hot-spotting in a Kafka consumer group

• Worked on Kafka offset monitoring and safer commit behavior during rebalances; learned the failure modes around ordering and duplicate processing in distributed consumers

• Built data validation checks to catch schema drift/type violations before downstream reporting pipelines

• Wrote Python utilities to automate ETL workflows into PostgreSQL for reporting