Jayash Bhandary
Full-Stack & AI Engineer · Bengaluru, India
findjayash@gmail.com · +91 91360 07794 · Portfolio · GitHub · LinkedIn
Summary
Full-Stack & AI Engineer with a year of production experience building LLM applications and autonomous agents end to end. Ships multi-provider RAG pipelines over vector databases and Model Context Protocol (MCP) servers that expose enterprise systems as reusable agent tooling — plus the React/Next.js front-ends, FastAPI services, and cloud deployments around them. Also works in Rust/WebAssembly and cross-platform desktop. Takes features from ambiguous requirement to deployed release.
Work Experience
- Shipped a multi-model RAG pipeline over SYNECA's 500K-document enterprise archive — ChromaDB retrieval with source-traced citations — cutting agent response time 5x.
- Engineered a multi-provider abstraction so the SIA agent switches between OpenAI, Claude, and Gemini at runtime, with vector-DB backends swappable the same way — no code changes.
- Designed and deployed an MCP server exposing SYNECA as 26 tools, consumable by any external agent or LLM client.
- Built an SEO-optimized Shopify storefront with a responsive, performance-tuned component library.
- Improved page load speed 30% with React.lazy code splitting, bringing the bounce rate down through faster loads and better UX.
- Launched 2went6ex.com, a non-contemporary fashion storefront, as sole developer — scope, design, build, testing, and deployment.
- Integrated a GraphQL Storefront API that delivered 8x faster product search and detail loading, on a responsive build verified across devices.
Projects
- Unified four connection paradigms behind one interface, letting users browse, edit, and query SQL, NoSQL, object storage, and HTTP APIs side by side in a single workspace.
- Engineered a Rust-to-WebAssembly core powering markdown editing, data conversion, task management, and image tooling fully offline, with optional Firebase sync.
- Built a Finder-style manager with an embedded Ollama chat panel, preview viewer, and workflow editor for chaining prompts against local files, keeping inference on-device.
- Developed a local agent executing app, file, and system commands from Telegram over an MCP tool layer, with authenticated real-time command handling.
Education
V.P.M's R. Z. Shah College, Mulund
VPM Junior College, Mulund
Little Flower High School, Thane
Skills
Languages: JavaScript, TypeScript, Python, Rust, C++, C#, Dart, SQL
Frontend: React.js, Next.js, Tailwind CSS, Framer Motion, Vite
Backend: Node.js, FastAPI, Express.js, Serverless, REST, GraphQL
AI / ML: PyTorch, Hugging Face, OpenAI API, Anthropic API, Gemini, Ollama, LangChain, RAG
AI Agents: MCP, LLM agents, multi-agent orchestration, prompt engineering, function calling, embeddings, vector search
Cloud & DevOps: AWS, GCP (Vertex AI), Docker, GitHub Actions, CI/CD, Terraform
Databases: PostgreSQL, MongoDB, Redis, ChromaDB
Systems: Rust, WebAssembly, Flutter, offline-first PWAs, cross-platform desktop
Certifications
Vertex AI Gemini API · GenAI Apps with Gemini & Streamlit · Google Cloud Essentials · Prompt Design in Vertex AI · Gemini Multimodality & Multimodal RAG · Neo4j Certified Professional · AI Apps with Gemini & Imagen · SQL to MongoDB Document Model · MongoDB Schema Design Patterns · RAG Course for Beginners