From production support to software architecture — I build systems, platforms, and products that survive the real world. I work across backend architecture, large-scale data systems, cloud platforms, AI engineering, and product development.
Leading engineering initiatives, core data platforms & cloud architecture.
Eliminated nested iterative scans and thread locks. Restructured memory chunking & B-Tree indices on live enterprise databases.
Support (2019) → Systems → Software → Scale → Architecture
Available for architectural advisory, deep performance bottlenecks, or collaborative product engineering. No recruiter middlemen or bureaucratic runarounds.
Grouped logically by discipline. Built on seven years of hands-on production delivery without arbitrary percentage meters.
Selected architectural blueprints and production artifacts across local-first software, event-driven data streaming, and scalable backend services.
Strength training without spreadsheet friction or cloud dependency.
Traditional workout tracking apps suffer from mandatory sign-in walls, sluggish cloud roundtrips in subterranean gyms, and feature clutter. SetKeep was engineered from first principles as an offline-first instrument: instant sub-16ms touch-to-disk write persistence, zero telemetry tracking, and reliable CRDT sync.
The Real Problem: Severe Out-Of-Memory (OOM) heap collapses when hundreds of concurrent enterprise users manipulated 500MB+ spreadsheet workbooks and triggered instantaneous redaction workflows.
The Architectural Fix: Replaced monolithic in-memory DOM representations with an asynchronous stream-based chunking pipeline backed by Redis Streams, enforcing strict bounded memory quotas per worker thread.
Iterating on local-first strength training companion, refining SQLite telemetry and offline cryptographic backup protocols.
Studying Hierarchical Navigable Small World (HNSW) graph indexing, memory alignment, and low-level systems programming in C.
Testing agentic workflow orchestration with stateful DAG guarantees, guardrailed execution gates, and CRDT convergence.
Re-reading Martin Kleppmann's chapters on distributed transaction commit protocols and real-world network partition handling.
Tactical learnings, edge cases, and runtime discoveries captured during production engineering.
Keyset pagination is often heralded over offset, but with composite indexes on (created_at, id) you must guard against nullability caveats. Ensure both columns are declared NOT NULL to avoid unindexed sequential scans when row tuples evaluate against null comparisons.
Enforcing deterministic JSON schemas at the inference boundary prevents cascading failures downstream. Coupling strict response schemas with circuit breakers and fallback regex parsers drops extraction crash loops to near zero during high-rate automated batch jobs.
Virtual threads eliminate carrier pinning only when synchronized blocks are systematically swapped for ReentrantLock. Legacy third-party drivers that execute I/O inside native synchronized methods will quietly monopolize OS carrier threads under load.
The publications that fundamentally reshaped my engineering mental models and architectural intuition.
The blueprint for understanding storage engines, consensus, replication lag, and fault tolerance in the wild.
The Real Problem: Dual-write drift between primary PostgreSQL relational tables and downstream search clusters, causing silent indexing state discrepancies and phantom reads.
The Architectural Fix: Tailed the Write-Ahead Log (WAL) directly using logical replication decoding slots into Kafka topics, pairing with transactional outbox mechanisms for guaranteed at-least-once ordering.
B-Trees vs LSM-trees demystified down to disk page layout, write amplification, and buffer managers.
Navigating technical leadership, driving organizational consensus, and setting multi-team technical strategy without direct authority.
Treating streaming logs as the central nervous system of enterprise platforms rather than ad-hoc message queues.
A tangible, chronological progression grounded in deep systems troubleshooting and scaling mission-critical platforms.
Security research and hands-on training.
Global infrastructure and enterprise engineering.
Digital engineering and enterprise platforms.
Healthcare AI and clinical intelligence platforms.