AI/ML Engineer · Agentic AI Architect · FinTech & Low-Latency Systems

Built and commercialized a local-first AI platform enabling sub-second LLM inference on consumer hardware, eliminating dependence on external cloud AI services and enabling privacy-preserving deployment. Architected and led end-to-end development of AI4All, a multi-tier production AI platform supporting local LLM inference, speech processing, semantic retrieval, and agent-based workflows. Delivered a privacy-preserving, on-device AI platform compliant with strict security standards. Shipped AI4All family of products to the Microsoft App Store allowing one-click installation. Architected an autonomous Agentic AI ecosystem with specialized agents for multi-step workflows. Designed high-performance RAG and semantic search systems with low-latency query execution. Engineered optimized inference pipelines delivering sub-second response times on commodity hardware. Built a modular, cross-platform AI framework in C++, and Flutter enabling deterministic, reproducible inference and extensible system composition.

Built and scaled a real-time financial analytics platform processing high-volume trading data, improving system latency and throughput by 20–30%. Designed and delivered a real-time streaming analytics platform using Kafka and Spark. Engineered low-latency data pipelines critical to live trading and risk decision systems. Led distributed engineering teams across New York, London, and India. Drove architecture improvements increasing throughput and system reliability under peak market hours.

Led modernization of legacy financial transaction platforms into cloud-ready microservices architecture, improving scalability and reducing system fragility. Architected migration of monolithic platforms to Spring Boot microservices. Designed and implemented high-throughput API integrations and backend services. Drove system stability improvements through refactoring and enhanced service orchestration. Mentored junior engineers and established coding standards.

Designed hardware-software co-optimization systems for heterogeneous compute environments. Developed performance-critical solutions combining hardware design (FPGA/Verilog) and low-level software (C++). Built custom compute architectures applicable to GPU/accelerator ML inference workloads.

Developed and maintained regulatory risk models (SVaR) supporting compliance reporting. Built and maintained Stressed Value-at-Risk models and automated reporting pipelines. Coordinated cross-functional delivery across risk, IT, and trading teams. Provided analytical support for model validation, governance, and regulatory review. Collaborated with global teams to align infrastructure with risk management priorities.