I build AI and data-heavy software with a backend-first mindset: reliable services, useful developer tools, auditable workflows, and interfaces that make complex systems easier to understand.

Currently based in California and working across applied AI, backend engineering, data systems, and product-focused technical tooling.

• AI applications that are practical, inspectable, and useful in real workflows
• Backend systems, APIs, data pipelines, and service reliability
• LLM tooling for retrieval, context optimization, local-first workflows, and serving tradeoffs
• Full-stack product experiences where technical depth and UX both matter

• Splitmaa: edge-device LLM mobile showcase for safe local function calling in expense workflows.
• TokenGate: context optimization for LLM apps with ranked, deduped, budgeted chunks and an audit trail.
• Beacon: local-first desktop RAG app for private document search and chat with local LLMs.
• Mini-AsymCache: visual simulator for exploring KV-cache policy tradeoffs in LLM serving.

• Applied AI systems and LLM infrastructure
• Backend architecture and API design
• Retrieval, RAG, and context engineering
• Data workflows, evaluation, and observability
• Performance analysis and system tradeoffs
• Developer tools and polished technical interfaces

Languages: C#/.NET, Python, TypeScript, JavaScript, SQL, Kotlin

Backend and data: APIs, data workflows, system design, local-first AI, RAG, evaluation

Frontend and product: React, Vite, Tailwind CSS, Recharts, mobile and desktop app experiences

Tools: Git, GitHub, Node.js, Vitest, Python tooling, automation-first workflows

• GitHub: github.com/Mario-Vishal
• Location: California, USA