Headline

Senior Backend Engineer | Node.js, TypeScript, GraphQL, MongoDB | Distributed Systems & Production AI (RAG, AWS Bedrock, LangChain)

About

I'm a backend engineer who builds the systems other teams build on — integration platforms, multi-tenant analytics, async job pipelines, and production AI search. At Delightree, a franchise-operations SaaS platform, I own the integrations, AI-search, and analytics subsystems across a GraphQL backend, an asynchronous operations service, and an AI analytics agent. I care about correctness under multi-tenancy, systems that degrade gracefully, and solving a problem once at the shared layer instead of N times downstream. On the backend side, I designed a reusable integration platform on Paragon (JWT auth brokering, a generic proxy client, a per-provider module/queue/route convention) that standardized five production integrations — HubSpot, QuickBooks, Zapier, Zenoti, and Read AI — with later providers reusing it unchanged. I built a multi-tenant location-analytics platform in Node.js and MongoDB with aggregation metric engines and time-series stores, using cron-driven per-entity Bull queue fan-out so one failing tenant can't take down the batch, feeding a validated pipeline into Google BigQuery. I also built a support-ticketing domain from scratch (17 GraphQL APIs, role-based permissions, MongoDB indexes), migrated a deeply nested data model to a flat collection with single-round-trip $lookup aggregations, and shipped a scoped-API-key partner ingestion API with hosted OpenAPI docs. Earlier, at Playo, I cut leaderboard read latency from 4.5s to 50ms with tiered Redis caching and brought geospatial search under a second across 600K+ venues. On the AI side, I built a multi-tenant conversational RAG engine over company SOPs using Weaviate hybrid search and AWS Bedrock (Claude/Titan), with session memory, rolling LLM summarization, citation-grounding guards, and strict per-tenant permission isolation across retrieval and storage. I own the LangChain/LangGraph tool-calling layer and instrument every query with Langfuse tracing and prompt management to monitor token cost, latency, and grounding behavior. Core stack: Node.js, TypeScript, GraphQL, MongoDB, Redis, Bull, AWS, BigQuery, Weaviate, and LangChain/LangGraph. Open to senior backend, platform, and AI-backend roles.

Location

Gurugram

AI stack

claude code

codex

langfuse

langgraph

langsmith

mcp

Work experience

Senior Software Engineer at Delightree, Apr 2026 - Present · 6 mos, - Architected a reusable third-party integration platform on Paragon — RS256 JWT auth brokering, a generic proxy client, and a per-provider module/queue/route convention — that standardized five production integrations (HubSpot, QuickBooks, Zapier, Zenoti, Read AI); later providers reused it unchanged. - Built a multi-tenant conversational RAG engine over company SOPs using Weaviate hybrid search and AWS Bedrock (Claude/Titan), with session memory, rolling LLM summarization, citation-grounding guards, and two-layer per-tenant permission isolation. - Developed the LangChain/LangGraph tool-calling layer and instrumented every query with Langfuse tracing and prompt management to monitor token cost, latency, and grounding-fallback behavior. - Designed and shipped an end-to-end Read AI meeting integration with rate-limit-aware incremental sync (429 backoff, quota short-circuit) and SHA-256 content hashing to protect human-edited records.

Software Engineer at Delightree, Jul 2025 - Mar 2026 · 9 mos, - Engineered a multi-tenant location-analytics platform (Node.js/MongoDB aggregation engines + time-series stores) with cron-driven per-entity Bull queue fan-out for failure isolation and a validated KPI pipeline into Google BigQuery. - Developed a support-ticketing (ITSM) domain from scratch — 17 GraphQL APIs, six MongoDB collections, configurable taxonomies with default-assignee routing, role-based permissions, activity logs, and compound indexes. - Migrated a location-launch task system from nested MongoDB documents to a flat collection, replacing two-query reads with single-round-trip $lookup aggregations plus dependency-based task unlocking and backfills. - Implemented a public partner ingestion API (scoped API-key auth, Zod validation, hashid-obfuscated IDs, org-hierarchy-aware provisioning) with hosted OpenAPI/ReDoc docs, and remediated a credential-exposure flaw in third-party log shipping.

Software Engineer at Playo, Sep 2024 - Jun 2025 · 10 mos, Bengaluru, Karnataka, India, ⚡ Engineered a real-time leaderboard system using a dense ranking algorithm and tiered Redis caching. Reduced response time from 4.5 seconds to just 50ms, directly enhancing user engagement during live tournaments. Enabled multi-dimensional filtering using MongoDB aggregation pipelines for dynamic views (e.g., sport-wise, friends-only). 🔁 Led a zero-downtime migration of Playo’s in-app chat to SendBird. Handled real-time API integrations and seamless user data transition, ensuring uninterrupted conversations for thousands of active users. 📍 Enhanced search experience across 600,000+ venues by integrating MongoDB Atlas Search with geospatial queries, cutting average query latency by 75%. This significantly boosted app usability and session retention. 📊 Launched a comprehensive user insight module to track lifecycle metrics like joining year, total games, and social connections. Migrated and normalized data across 3 MongoDB collections within 50 minutes while ensuring zero data loss—powering new retention and loyalty features. ⭐ Impact: Drastically improved performance, personalization, and data reliability across Playo’s core features, directly contributing to enhanced user engagement and retention.

Software Developer at Teal India, May 2023 - Aug 2024 · 1 yr 4 mos, 📝 Developed an end-to-end automated legal reporting engine that generated 350+ due diligence reports daily by parsing structured and unstructured legal data—cutting manual effort by 80% and revolutionizing legal ops. 📡 Built real-time analytics dashboards with WebSocket support, introducing sticky-sessions for horizontal scaling—empowering teams with live, interactive insights across the application. 📚 Created a custom PDF highlighter library with OCR and smart page navigation tailored for legal workflows, leading to a 3x faster document review process for analysts and lawyers. 📈 Engineered interactive dashboard components using custom React hooks and D3-style visualizations, boosting user interaction time by 25% and making data stories more intuitive. ⭐ Impact: Accelerated legal operations and significantly elevated user experience through intelligent automation, scalable architecture, and real-time systems.

Software Developer Intern at Teal India, Feb 2023 - Apr 2023 · 3 mos, 🏗️ Re-architected core backend modules, modularizing services and reducing compute latency by 50%. Migrated key components to Next.js SSR to improve performance and SEO for client-facing tools.

DevOps Engineer at Recruit CRM, Oct 2022 - Jan 2023 · 4 mos, New Jersey, United States, 🌐 Increased platform reach by 75% through implementation of NAT Gateway Elastic IP whitelisting, enabling recruiters in restricted regions to access cloud-hosted services seamlessly. 🛠️ Optimized system configurations to streamline network routing, contributing to more stable outbound communications and improved global access latency. ⭐ Impact: Significantly improved platform accessibility and system performance, resulting in smoother international operations and reduced support overhead.

Back End Developer at Duckcart, Aug 2022 - Sep 2022 · 2 mos, Noida, Uttar Pradesh, India, 🧱 Led backend development from the ground up, collaborating closely with cross-functional teams to create a modular and high-throughput system capable of scaling with increasing user demand. 🔧 Advocated for engineering best practices and facilitated smoother developer onboarding by standardizing codebases and establishing documentation guidelines. ⭐ Impact: Delivered a reliable, scalable backend infrastructure that served as the foundation for the company’s early growth and stability.

Education

Bachelor of Technology - BTech, Computer Science and Engineering, Dr. A.P.J. Abdul Kalam Technical University

Intermediate, kendriya vidyalaya

Public links

amantyagi.me: https://amantyagi.me/