NocodeNoproblem

Project Concept

Two builders who believe the next interface isn’t designed — it’s described. We’re here to turn plain English into live, interactive UI using agentic frameworks and Claude. No Figma. No templates. No drag and drop. Just intent, and what comes alive from it.

Entry

Status: Submitted

Last saved: February 21 at 6:23 PM EST

View Entry

Team Roster

You must be registered for the event to view the team message board.

Akanksha Kushwaha Team Lead RSVP Approved

AI Engineer Volunteer at American Red Cross

Architected and built the full system end-to-end. Designed and implemented the Python A2A agent using the A2A SDK, including the UIGeneratorExecutor, schema validation loop, and auto-retry mechanism. Engineered the conversational memory system that passes full conversation history through A2A message metadata, enabling Claude to make surgical refinements without full rebuilds. Built the Next.js 15 frontend including the blank canvas, version timeline, preview modal, and refine bar. Integrated the A2UI v0.8 protocol and Lit web components renderer. Wrote the Claude system prompt with A2UI schema constraints and component catalog. Leveraged Anthropic's claude-sonnet-4-5 via the Anthropic SDK as the core reasoning engine driving all UI generation.

I’m an AI Engineer focused on building production-grade, agentic AI systems, especially in enterprise and regulated environments. I started my career in data engineering, working on large-scale cloud migrations and ETL systems, which gave me a strong foundation in how data moves and scales. Over time, my focus shifted toward the modeling layer — understanding how data is transformed into representations that drive predictions and decisions. Today, my work centers on retrieval-augmented architectures, agent routing, guardrails, and evaluation pipelines. I’m particularly interested in modeling behavior, evaluation trade-offs, and how AI systems generalize — not just in isolated benchmarks, but in real-world deployments. I spend a lot of time thinking about failure modes, observability, and

I’m particularly interested in advancing prod-grade agentic AI systems, with a focus on evaluation, observability, & system-level reliability. I’m eager to explore how multi-step AI workflows can be measured and continuously improved through structured feedback loops. I’m also interested in LLMOps, AI infrastructure, and secure orchestration frameworks that enable enterprise-scale deployment. I enjoy connecting with engineers and product teams working on AI evaluation, behavior, and deployment.

● Designed and deployed a guardrail-enabled, multi-agent AI assistant simulating American Express compliance workflows using AWS Bedrock, Strands Agents, and FAISS. Integrated PDF policy documents and Notion-based FAQs using Titan embeddings within a hybrid RAG pipeline, achieving 90% semantic alignment, improving query response accuracy 3×, and enabling live guardrail monitoring through Weights & Biases dashboards.

Kunal Singh RSVP Approved

Data Governance Manager at American Express