Shadrach – AWS, React, Typescript

Built a multi-tenant AI agent system that automates sales and customer support for SME, It is multi-channel agent system that works across WhatsApp and Instagram. Handles customer interaction, queries, scheduling and payments.

• Architecture and development of a multi-tenant customer support AI agent harness with support for multi-channel (whatsapp/Ig), connectors for dynamic tool integrations, Hybrid Rag with Multimodal vector and BM25 retrieval, Human-in-the-loop (HITL) escalation support.
• Designed a dynamic MCP based tool orchestration layer supporting 10+ reusable business capabilities (scheduling, catalog lookup, booking, payments) eliminating what would otherwise be custom integrations for every new client.
• Instrumented end-to-end conversation tracing across every tenant, reducing debugging time for production agent failures from hours to minutes.
• Built an automated evaluation and learning pipeline that reduced manual prompt-tuning effort by 50%+ while maintaining response quality as conversation volume increased.
• Partnered directly with the founding team to turn a fast-changing product surface into reliable agent behavior under live customer traffic, with no dedicated ops team backing it up.

An AI-powered research assistant platform built on indexed peer-reviewed literature, enabling real-time collaboration between researchers.

• Developed a critical data ETL pipeline for Indexing 250M+ peer-reviewed research records from OpenAlex dataset on Elasticsearch node, giving the AI research assistant citation-backed coverage across the full scientific corpus rather than a narrow slice of it.
• Built real-time collaboration infrastructure (Cloudflare Durable Objects, PartyKit, CRDT/Automerge, Postgres) supportingconflict-free concurrent editing of research artifacts, contributing to 200% user growth within weeks of release.
• Owned Kubernetes deployment, monitoring, and production operations across AWS, reducing mean time to detect production issues through centralized observability with Prometheus and Grafana.
• Shipped the research assistant's core API and multiple product features powering citation-backed search across 250M+ scientific records
• Contributed core smart-contract work (Solidity, Foundry) to the Desci Codex protocol, underpinning the trust and provenanc guarantees behind every published research object.
• Led UI engineering for Nodes protocol web app, architectural setup, blockchain integration (wallets, ethers.js, Wagmi, etc.), performance optimization and avoiding SSR pitfalls for modern web3 apps.

A DeFi protocol and DAO ecosystem with staking, liquidity management, and analytics.

• Designed and deployed multiple production Solidity smart contracts;
• Architected serverless backend infrastructure using AWS Lambda + DynamoDB;
• Built staking dashboards and analytics web apps with Next.js + TypeScript;
• Implemented subgraph indexing for blockchain data querying;
• Led end-to-end feature development across smart contracts, backend, and frontend.

A multi-chain crypto donation platform enabling nonprofits to receive donations across Bitcoin, Etheruem, EVM-compatible L2 chains and alternative chain.

• Led development of the multi-chain donation UI using React, TypeScript, ethers.js, wagmi;
• Provided blockchain Integrations across multiple EVM-compatible chains;
• Contributed to Cosmos Rust smart contracts;
• Implemented unit, integration, and E2E tests;
• Designed wallet connection and transaction UX flows.
• Cross-chain event indexing on MongoDb and development of internal analytics dashboard

Book-IT is a Hotel booking platform designed with clean layered architecture, role-based access control, and production-ready engineering practices. The system allows users to browse hotels, book rooms, and manage reservations, while admins manage inventory and operations.

• Demonstrates end-to-end ownership of a SaaS-style booking system;
• Production-ready authentication & access control system;
• Automated testing integrated into development workflow.

• Designed and implemented a layered REST API architecture;
• Automated test suites using Jest + Supertest for service and domain layers;
• Enforced code quality via Husky pre-commit and pre-push hooks.