Built a compiler from scratch with multi-level IR (HIR->MIR), SSA construction/deconstruction, dominator analysis, and data-flow analysis, featuring GVN, LICM, DCE, DBE, and TCO
Implemented deterministic CFG-to-structured WebAssembly lowering via Ramsey (2022) for reducible control-flow graphs
Built a C-like programming language from scratch with first-class functions, pointers, stack frame management, abstract data types, and basic optimizations such as constant folding, targeting x86 AT&T assembly
Developed custom testing and benchmarking frameworks in Python for correctness verification and performance evaluation against C/C++
Built a Bitcoin-inspired blockchain platform with proof-of-work consensus, libp2p-based P2P networking, dynamic difficulty adjustment, longest-chain fork resolution, and gas-metered smart contracts executed on a custom stack-based virtual machine (AVM)
Developed Marigold, a high-level smart contract language with first-class functions, static typing, persistent storage, and a full compiler targeting AVM bytecode
Lead architecture and infrastructure for multi-cloud deployments (AWS, GCP, Yandex), defining service boundaries, scaling strategies, and regional hosting for AI inference
Built scalable Python microservices and Java Spring Boot multi-tenant backend, replacing legacy diagnostics engine with low-latency AI inference, achieving ~40% latency reduction
Designed AWS SQS-based queue system to scale AI services horizontally, improve fault tolerance, and support enterprise B2B integrations
Oversaw technical scoping for enterprise clients, solution specifications, infrastructure cost optimization (FinOps), and cloud contract negotiations
Managed a cross-functional team of 5 engineers, set technical standards, and coordinated model integration pipelines