Full Stack Developer · Finanshels Feb 2025 – Apr 2026 | Remote, India

• Spearheaded development of a comprehensive Practice Management platform (Next.js + NestJS) automating full-cycle organisational workflows — bridging sales pipelines, project management, and accounting systems — with PostgreSQL schemas and DynamoDB Streams optimised for high-throughput billing workflows.
• Architected a production LLM-powered AI Bookkeeper using a custom multi-agent orchestrator (CrewAI + LangChain); designed LLM-ready APIs and RAG pipelines with vector retrieval, reducing bookkeeping TAT by over 90% (weeks to under one day).
• Implemented end-to-end LLM evaluation pipelines tracking retrieval precision, tool-calling accuracy, and hallucination rates in live environments — enabling continuous quality improvement on AI workflows.
• Built and deployed a secure, centralised client portal integrating unified client profiles, timesheet tracking, task management, and automated report delivery — significantly improving operational visibility and client support delivery.
• Built third-party integrations with accounting platforms, CRMs, and billing systems — designing normalised data pipelines and webhook-driven sync layers to keep interconnected systems in real-time consistency.
• Designed event-driven microservices on AWS (Lambda, DynamoDB Streams, SQS) ensuring fault tolerance, high availability, and low-latency state propagation across interconnected financial data pipelines.
• Integrated MCP (Model Context Protocol) servers for secure, standardised tool access across agent workflows — enabling scalable inter-agent communication and external API orchestration.
• Directed engineering operations in the absence of the VP of Engineering — managed CI/CD deployment pipelines, led daily stand-ups, and drove technical decision-making during high-level strategic planning.
• Led incident management, root cause analysis, and observability setup; championed API-first, LLM-ready design principles and mentored junior engineers through code reviews and design discussions.

Tech Stack: Python, Next.js, NestJS, TypeScript, PostgreSQL, DynamoDB, AWS (Lambda, SQS, S3), CrewAI, LangChain, Docker, Redis, pgvector

Python Developer (Backend) · The Good Glamm Group May 2024 – Feb 2025 | Mumbai, India

• Migrated legacy monoliths to event-driven microservices on Docker/AWS; designed async SQS/SNS pipelines for real-time inventory and order state propagation serving millions of concurrent users.
• Built and optimised FastAPI / Fastify backend services; improved p99 latency by 35% on high-traffic read paths through targeted SQL and NoSQL query optimisation (PostgreSQL, MongoDB).
• Integrated AI-driven semantic search using embeddings and vector retrieval to boost product discovery relevance; improved recommendation quality measurably via A/B testing.
• Streamlined CI/CD pipelines (GitHub Actions) reducing release cycles by 40%, enabling safe, rapid feature rollouts with automated test coverage gates.

Tech Stack: Python, FastAPI, Node.js (Fastify), MongoDB, PostgreSQL, Redis, AWS (SQS/SNS, EC2), Docker, LLM APIs

loopvoice.ai — AI-Powered Voice Automation Platform for Shopify | Live Product | Built in 1.5 months

• Built and shipped a fully live AI voice automation platform for Shopify stores in 1.5 months using AI-assisted tooling (Cursor, Claude Code) — integrating voice AI APIs (STT/TTS) to automate abandoned cart recovery, order follow-ups, and customer outreach via automated voice calls, email, and SMS.
• Designed a DAG-based visual flow builder — modelled flows as Directed Acyclic Graphs (DAGs) to enable merchants to create event-triggered automation (abandoned checkout, order placed, etc.) with configurable wait nodes and branching logic — integrated Redis-based job scheduling for reliable timed execution.
• Built custom AI voice agents with RAG-powered knowledge bases (ingesting store websites and policies) to handle real customer queries — and a campaign feature for batch-calling segmented customer lists with configurable call frequency and rate controls.
• Integrated Telnyx for in-platform phone number provisioning across global regions; full-stack built on Next.js, PostgreSQL, GCP, and Redis.

Next.js, PostgreSQL, GCP, Redis

bot9.ai — Enterprise Customer Support Chatbot Platform | Used by RentoMojo PAN India

• Built a production-grade enterprise chatbot platform currently deployed for RentoMojo across India — supports configurable bot logic, team member management, and omnichannel integrations (Slack, Discord, Freshchat, WhatsApp).
• Enabled direct API integration with enterprise production databases — allowing the bot to act on live data (orders, rentals, accounts) rather than static knowledge; powered by Typesense (vector search) for semantic retrieval and built on Node.js, React, and PostgreSQL.

Node.js, React, PostgreSQL

OpenAGI — Open-Source Autonomous Agent Framework | Feb 2024

• Built core backend infrastructure for an open-source autonomous agent framework; implemented MCP servers enabling secure, scalable tool integration and inter-agent communication across distributed workflows.
• Engineered custom planning algorithms, memory-augmented state machines, and self-correcting execution pipelines with fault recovery — applied Control Plane architecture to enable long-running agentic workflows.

LegalEase — Compliance & Document Automation Platform for MSMEs | Nov 2023

• Designed a scalable FastAPI backend with automated OCR ingestion pipelines, RESTful microservices, and semantic document retrieval using vector embeddings and pgvector.
• Implemented secure data storage and retrieval with AWS S3 and IAM role-based access controls, ensuring enterprise-grade data privacy and encryption compliance.