Sankalp Gupta

Skills

Experience

• Software Developer at Teamie

  2024-02-01 – Present

  Agentic AI Platform and LMS integration. Built Enterprise-grade Multi Agent AI systems using LangChain, LangGraph, and OpenAI models for autonomous workflows. Developed Agentic AI RAG workflows multi step planning, tool orchestration, and evaluation loops to automate complex research tasks, reducing latency and 100% factual accuracy tailored to the user's specific context and diverse learning objectives. Implemented Human-in-the-Loop review workflows using LangGraph interrupts for instructor validation of AI-generated Quiz and Grading Assessments. Developed dynamic tool-selection mechanisms enabling AI agents to autonomously determine best tool invocation paths for answering user queries. Setup the PostgreSQL checkpointer using PostgresSaver to add persistent memory to the AI Agent. This saves conversation threads and prevents data loss, allowing interactions to resume instantly after pauses or crashes. Built a centralized logging and monitoring system using Winston, Morgan, and custom middleware for application, HTTP, and business-level request tracking. Implemented structured request logging by capturing metadata such as clientId, IP address, route, headers, and API method into the database for auditability and debugging. Optimized API performance using compression middleware (gzip) to reduce response payload size, bandwidth usage, and improve response times for AI workflows. Managed Node.js production deployment using PM2 with environment-based configuration, automatic process restart, monitoring, and log management. Enforced consistent coding standards and code quality using ESLint with customized rules, improving maintainability and reducing style/quality issues across the codebase. Created microservices featuring Node.js file-parsing utilities for document text extraction and an OpenAI Whisper-powered pipeline for automated audio and video speech-to-text generation. Deployed Docker-containerized backends with Milvus Vector DB and PM2-managed frontends on AWS EC2 infrastructure. Designed high-concurrency AI backend systems optimized for low-latency production traffic and large-scale concurrent users. Implemented secure Google APIs authentication using encrypted service accounts, domain-wide delegation, and dynamic user impersonation for permission-aware content access and processing. Integrated LMS authentication APIs into the AI Agent framework, enabling clients to securely access tenant-specific AI services.

Projects

• Instructor Agent (ZilAgent)

  2025-12-01

  Built this LMS-integrated multi-agent AI system which have multiple Agents as Grading Assessment, Personalized Quiz Generation, Student Feedback, and Learning Path Adaptation. Integrated with LMS platforms to track student profiles, student performance, and grades and ease the work for instructor for quiz generation, assessment grading, feedback creation, and personalized learning adaptation.

  LangChainLangGraphNode.jsPostgreSQL

• AI Assistant

  2025-07-01 – 2025-11-01

  Built an AI Assistant platform using OpenAI Assistant API with multimodal ingestion support for files, PDFs, URLs, audio, video, and images. Managed assistant creation, updates, conversations, and deletion via OpenAI APIs.

  Node.jsExpress.jsOpenAI Assistant APIJWTAngularJs

Career preferences

Useful for recruiters: location, job types, salary range, and availability.

Posts