Hi, I'm Sarthak.
Full Stack AI Developer.
Building applied AI systems and production backend applications with a focus on reliable backend services, AI workflows, guardrails, and observability.
I'm a backend focused full stack developer building applied AI systems.
I build production web applications and backend services using Node.js, TypeScript, Python, FastAPI, REST APIs, and databases. My work includes authentication, role-based access control, data-driven applications, asynchronous workflows, and cloud deployment.
I also build applied AI systems with LangGraph, RAG, MCP, and LLM-based workflows, with a focus on guardrails, human-in-the-loop approval, evaluation, reliability, and observability.
I like systems that are reliable and easy to reason about: clear API contracts, resilient background processing, well-defined service boundaries, and AI workflows that are observable and safe to operate.
Nov 2025 — Present
Web Developer
Lebeda Marcom and Creative Studio LLP
Bengaluru, India · Onsite
- Built and delivered 5+ client websites end-to-end using Node.js (v20+), React.js (v19), Next.js (v16), and Tailwind CSS (v4), handling development, integration, and deployment.
- Developed a full-stack finance platform using TypeScript, React, Express, and PostgreSQL, implementing REST APIs, JWT authentication, RBAC, Zod validation, and Puppeteer-based PDF generation for financial, employee, order, and document workflows.
- Developed Node.js (v22) backend workflows for a live client platform, supporting booking approvals, user registration, automated email notifications, and MongoDB data persistence with Google Sheets synchronization.
Project 01
ShopSense - E-Commerce Customer Support System
- Built a multi-agent LangGraph workflow with ticket triage, escalation, MCP-based order tools, and human-in-the-loop approval for high-risk actions.
- Implemented hybrid RAG using Qdrant dense retrieval and BM25-style retrieval, combined with Reciprocal Rank Fusion (RRF) and cross-encoder reranking.
- Added groundedness checks, LLM-as-judge validation, prompt-injection detection, and guardrails around order actions such as refunds and returns.
- Added reliable agent execution with retries, circuit breaking, and Langfuse observability for tracing and monitoring AI workflows.
Python · FastAPI · LangGraph · RAG · Qdrant · MCP · LiteLLM · Langfuse · React · TypeScript
Project 02
Log Drain - Log Observability Platform
- Built a Redis-buffered asynchronous log ingestion pipeline with 202 responses, background workers, reliable PostgreSQL inserts, and worker coordination.
- Implemented hybrid keyword and semantic search using PostgreSQL full-text search, pgvector, and Reciprocal Rank Fusion (RRF).
- Built a LangGraph tool-calling agent with tools for log search, statistics, anomaly detection, and service lookup, with real-time SSE response streaming.
- Built a FastAPI anomaly-detection service using IsolationForest and deployed the platform to AWS EC2 with Docker Compose, alongside project-scoped API-key authentication.
Node.js · Express · TypeScript · PostgreSQL · Redis · pgvector · LangGraph · FastAPI · React · AWS EC2 · Docker
Programming Languages
JavaScript · TypeScript · Python · C++Frameworks & Technologies
React.js · Next.js · Node.js · Express.js · FastAPI · REST APIs · JWT · RBAC · LangGraph · RAG · MCP · PostgreSQL · MongoDB · Redis · Vector DatabaseTools & Platforms
Git · GitHub · Docker · AWS EC2 · Postman · VS CodeApplied AI Professional Certification Program
Indian Institute of Technology Hyderabad
August 2026
LINKBachelor of Engineering
Computer Science & Engineering