Database & Systems Engineer specializing in low-level systems programming with focus on performance-critical applications. I build high-performance data tools, database engines, network services, and system utilities from scratch using Zig, Rust, and Go. My work emphasizes zero-GC architectures, memory safety, and ultra-low latency designs that achieve orders of magnitude improvements over traditional solutions.

Currently building Glacier - a pure Zig OLAP query engine that processes Apache Iceberg tables and Parquet files with ~8 MB binary size, 5-50 ms cold start, and zero garbage collection pauses.

💻 Technology	🚀 Projects
	Coming soon...
	Coming soon...
	Coming soon...

OLAP & OLTP Database Systems

• SQL query engines: parser design, optimizer, execution planner, expression evaluation
• Aggregate and window functions with multi-dimensional grouping (ROLLUP, CUBE, GROUPING SETS)
• Query optimization: predicate pushdown, column/file pruning, join reordering, cost-based optimization
• Indexing structures: B-Tree, B+Tree, B*Tree, Hash indexes, Bitmap indexes
• Storage internals: Write-Ahead Logging (WAL), Log-Structured Merge Trees (LSM-Tree), buffer pool management
• MOLAP storage engines with dimensional hierarchies and cube materialization
• ROLAP systems and data warehousing schemas (Star, Snowflake, Galaxy)
• Transaction management: MVCC, isolation levels, distributed transactions, two-phase commit

Low-Level Systems Programming

• Memory management: custom allocators, arena allocation, memory pools, zero-copy architectures
• Concurrency primitives: lock-free data structures, atomic operations, thread synchronization
• Performance optimization: SIMD vectorization, cache-friendly algorithms, CPU profiling
• Network programming: TCP/IP socket handling, HTTP/HTTPS clients, protocol implementation
• Compression algorithms: Snappy, LZ4, Zstd, dictionary encoding, run-length encoding
• Binary protocols: Thrift, Protocol Buffers, Apache Arrow, custom wire formats
• File I/O optimization: mmap, async I/O, direct I/O, buffered streaming

Modernization & Performance Engineering

• Legacy system rewrites: migrating monolithic JVM applications to lightweight native binaries
• Cost reduction: 10x-100x reduction in cloud compute costs through efficient resource utilization
• Latency optimization: sub-millisecond query response times, eliminating GC pauses
• Memory footprint reduction: from GB-scale heap usage to MB-scale deterministic allocation
• Cold start elimination: millisecond startup times vs multi-second JVM initialization
• Deployment simplification: single static binary vs complex dependency management (JARs, runtimes)
• Horizontal scalability: stateless architectures, shared-nothing designs, data partitioning strategies

Zero Hard-Coding - Everything generic and dynamic. No schemas baked into code. Runtime type resolution for all data formats.

Memory Safety - No leaks, no undefined behavior. GPA-verified allocations. Arena-based management with deterministic cleanup.

Performance First - Ultra-low latency targets. Minimal memory footprint. Zero garbage collection. Built for production workloads.