Vishvendra Tomar
A Technical Overview

Abstract

Backend engineer specialising in distributed systems, async task pipelines, and LLM integrations. Built and debugged production systems at scale — from observability pipelines and Celery task queues to AI-powered developer tools with RAG, vector search, and structured LLM output. Experienced integrating OpenAI and Anthropic APIs across multiple platforms, with focus on production reliability, log-driven debugging, and automation agents.

1. Introduction

As backend systems grow in complexity — distributed architectures, async pipelines, and LLM integrations becoming table stakes — the need for engineers who can build, debug, and ship these systems at production scale has intensified. This paper presents Vishvendra Tomar, a Backend Engineer with a track record of delivering production-grade systems across AI tooling, distributed infrastructure, and enterprise platforms.

Research Interests: Distributed Systems, LLM Integrations, RAG Systems, Production Debugging, AI Tooling, Async Pipelines.

Current Position: Software Engineer 2 at GeekyAnts, focusing on backend architecture, LLM integrations, and distributed systems.

2. Methodology & Technical Stack

Our approach leverages a diverse technical stack across multiple domains:

3. Experimental Implementations

We present 4 notable implementations demonstrating practical applications of backend and AI engineering:

3.1. myJobDrop — Workforce Management Platform [1]

Production platform scaling to 300K+ users. Led full migration from 50+ Firebase Cloud Functions to modular FastAPI backend (13 modules, 46-table PostgreSQL schema with RLS). Built ACID-compliant Celery pipelines with exponential backoff, dead-letter queues, and Prometheus+Loki-driven debugging that eliminated 13 atomicity violations. Built stateless domain engines: wage calculator, 15-rule shift validator, automated invoicing. Integrated GPT-4o with structured output parsing for 93-category job classification.

Technologies: Python, FastAPI, Celery, PostgreSQL, Redis, Docker, Railway, OpenAI API, Pydantic, Prometheus

3.2. AntAgents — Enterprise Workflow Automation [2]

Enterprise platform with visual drag-and-drop process builder (React Flow). LLM-powered AI agent orchestration as pluggable workflow nodes. Real-time process execution engine with state management, pause/resume, trace ID-based debugging. Event-driven via Kafka + RabbitMQ. Full observability: Prometheus, Grafana, Loki across 11 containerized services.

Technologies: NestJS, Next.js, React Flow, MySQL, Prisma, Kafka, RabbitMQ, Redis, BullMQ, Docker

3.3. workspace.build — AI-Powered Dev Platform [3]

AI developer platform with 106+ REST APIs across 22 modules. Claude + GPT-4 integration with WebSocket streaming and structured LLM output parsing for 3-phase AI brainstorming. Temporal workflows for 5-stage project creation. Kubernetes-based isolated dev workspaces per user (VS Code Server + Terminal).

Technologies: NestJS, Next.js, PostgreSQL, Temporal, Redis, Kubernetes, Stripe, GitLab API, Anthropic Claude, OpenAI

3.4. Code Analyzer — AI Code Analysis Tool [4]

Claude AI-powered code review with risk assessment and performance scoring. Semantic code search via ChromaDB + Sentence-Transformers (RAG pipeline). GitHub integration for commit history, contributor metrics, productivity visualizations, cycle time breakdown, activity heatmaps.

Technologies: NestJS, Python Flask, PostgreSQL, ChromaDB, Anthropic Claude API, Sentence-Transformers, GitHub API

4. Results & Impact

Key achievements from recent work:

• Promoted twice in under 2 years (Intern → SE → SE2)
• Migrated production platform to serve 300K+ users with zero downtime
• Eliminated 13 atomicity violations through log-driven production debugging
• Built RAG pipelines and LLM integrations across 3 production platforms
• Full observability stack (Prometheus, Grafana, Loki, Jaeger) across all systems

5. Conclusion & Future Work

This work demonstrates comprehensive expertise in backend engineering, with proven ability to deliver production-grade distributed systems and AI integrations. Currently pursuing advanced applications in LLM tooling, RAG architectures, and scalable backend infrastructure. Open to collaboration on challenging backend + AI engineering projects.