Abhimanyu Sikarwar

• Build and maintain the React and Next.js UI for Kodo's spend platform in TypeScript, on a headless Form.IO architecture that lets finance teams define approval forms and workflows without a release. Delivered procurement in 6 months against a 2-year plan, then led Mensa Brands' migration from Zoho onto it.
• Use Claude Code, Cursor, Copilot, and Codex daily to build and prototype components, and set the review standards the team applies to AI-generated code before it merges.
• Built the streaming chat interface for an agentic AI reporting assistant (LangChain, LangGraph): tokens render as they arrive, long report answers stay interactive, and finance teams query invoices, reimbursements, and transactions in natural language.
• Fixed the load and interaction bottlenecks on the dashboard with route-level code splitting, lazy loading of heavy report views, and CDN caching, tracking LCP, INP, and CLS as the measure.
• Led the corporate card platform end to end: UI automation across issuance, limits, and expense capture lifted program efficiency by 40% and cut expense management costs by 25%; fraud detection and risk analysis in the card flows brought fraudulent transactions down 30%.
• Raised frontend test coverage from 15% to 80% and cut the production bug rate by 60% through code review standards and CI gates, working across a team of 5 spanning frontend, backend, and QA in weekly sprint planning and design critiques.

• Owned the shared component library across the product line and enforced consistency on every surface built from it, which sped up delivery on each project; led 3 frontend engineers.
• Translated Figma specs into pixel-accurate components, then took the UI to WCAG 2.1 AA: audited against the standard and fixed keyboard navigation, focus management, ARIA semantics, and contrast.
• Led the UI architecture for AUSIS, a machine-learning underwriting platform deployed at Aditya Birla Sun Life Insurance; the interface rendered real-time model risk scores and cut manual underwriting time by 40%.
• Built Insta Insurance, a digital platform for instant term life purchase: onboarding got 60% shorter, policy issuance ran 75% faster, and the product reached 15,000+ customers.
• Worked directly with backend and ML engineers to wire Python model APIs into the frontend for Wipro Intelligent Underwriter, an AI risk assessment platform built in partnership with Artivatic.
• Built a vehicle insurance renewals CRM where lead management joins customer and vehicle data to automate renewals, NCB-based premium adjustments, and dynamic discounting.

• Built Houzzcart, a home decor e-commerce platform, in Angular 11 and Node.js; delivered in 4 months and improved UI performance by 30%.
• Cart abandonment sat at 45% at checkout; put Stripe and Paytm behind one checkout UI with error recovery flows, which raised conversion by 25% and cut payment failures by 40%.
• Page load reached 3 seconds under traffic spikes; added Redis caching, query tuning, and CDN delivery to bring it under 1.5 seconds and triple capacity.
• Built the EdTerra Edventures learning platform with blogs, podcasts, quizzes, and location-based learning, which lifted student engagement by 40%.

React Native · Expo · TypeScript · React · Node.js · MongoDB · Cloudflare Workers

• Shipped an Android app in React Native and Expo plus a React web app in six weeks; the app is live on Google Play.
• Built the chat UI over Anthropic and OpenAI APIs with streaming responses and generative content rendered as it arrives, plus a document capture flow where photographed lab reports and ECGs come back transcribed for review.
• Wrote clinical-mcp, a TypeScript MCP server with 60+ tools covering ICD-10/11, LOINC, and RxNorm lookups, drug interaction checks, and 35 deterministic clinical calculators, deployed on Cloudflare Workers.

React · TypeScript · WebSocket

• Built a real-time analytics dashboard in React and TypeScript over a WebSocket feed, which cut decision-making time by 35%.
• Added filtering with saveable presets so traders keep a view per strategy; daily active engagement rose 45%.
• Designed a responsive S&P 500 heatmap with sector-based rendering so the whole index reads on one screen.

MCP · LangChain · LangGraph · OpenAI API · Claude API · Python

• A diet-planning agent builds weekly meal plans from health preferences, then a procurement agent breaks them into ingredient lists and places the orders through store APIs, with MCP handing context between ChatGPT and Claude across both steps.