Initializing environment
...
Setting up backend...
Python Backend Engineer specializing in Search Systems, Information Retrieval, and Scalable Applications.
This is what I build — live search relevance & backend deployment, visualized.
Core Engineering Pillars
Search & Retrieval Systems
Specialized in Azure AI Search, semantic ranking, scoring profiles, relevance tuning, index design, fuzzy matching, and low-latency search API architectures.
Backend & Distributed Systems
Architecting scalable Python & Django services, PostgreSQL data layers, Redis caching patterns, and asynchronous Celery task pipelines.
AI Applications & Agents
Designing Retrieval-Augmented Generation (RAG) systems, vector similarity retrieval (pgvector/HNSW), and Python-based AI agent workflows.
Professional Experience
Application Programmer
Tata Consultancy Services
Working on search-driven applications and information retrieval systems using Azure AI Search.
Contributing to search relevance tuning, semantic ranking, fuzzy matching, search suggestions, index design, query optimization, and backend application development.
Exploring AI agent architectures and retrieval-augmented generation (RAG) systems built with Python.
- Engineered search relevance pipelines using custom scoring profiles, field boosting, and semantic re-rankers
- Designed search indexes, fuzzy search configurations, and search suggestion endpoints
- Optimized query execution plans, payload serialization, and database access patterns on PostgreSQL
Built for Engineers.
Publishing libraries, rendering engines, and automation frameworks on PyPI and GitHub with strict typing and zero browser runtime bloat.
ProfileForge
Mathematical SVG Generation Engine for Developer Profiles
A developer tool and open-source library distributed on PyPI that renders dynamic, theme-aware GitHub profile dashboards without any external headless browser runtime (like Puppeteer or Chromium).
- Strict 6-layer unidirectional architecture: Data → Fetch → Parse → Layout → Render → Output
- Deterministic two-pass layout calculation engine for SVG element bounding boxes
- Automated upstream API snapshot locking via GitHub Actions to prevent breaking schema drift
- Configurable theme parser with CSS variable injection and vector glyph rendering
PglRobot
High-Throughput Hybrid Telegram Moderation Engine
An asynchronous event-driven moderation and automation engine built with a hybrid MTProto architecture, combining Bot API convenience with client-level protocol capabilities.
- Dual-engine architecture: Aiogram 3 for Bot API command handling + Telethon for MTProto deep inspection
- Non-blocking I/O throughout the execution pipeline using asyncpg and SQLAlchemy 2.0 async sessions
- Decentralized federation network for real-time cross-group synchronization of ban lists and heuristics
- Event-driven listener pipeline with strict Pydantic data schemas and rate-limited dispatchers
Production Platforms
Selected Case Studies
Attendate
Flagship SystemEvent Discovery & Management Platform
Overview
A full-featured backend platform for event discovery, ticketing, and attendance management. Built with Django as the core framework, backed by PostgreSQL for relational data, Redis for caching, and Celery for asynchronous workflows including payment processing and notification delivery.
Engineering Focus
- Multi-layer Django architecture with clean separation of concerns across apps, services, and repositories
- Recommendation engine using collaborative filtering to surface relevant events based on user behavior
- Asynchronous payment workflow with Celery — decoupled from the request cycle with retry logic and idempotency
- Redis-backed session management and event feed caching to reduce database query pressure
- JWT-based authentication and role-based authorization for organizers, attendees, and administrators
- Background processing for email notifications, ticket generation, and analytics aggregation
URL Shortener
High-Performance Link Shortening Service
Overview
A URL shortening service with real-time click analytics, rate limiting, and intelligent request routing. Designed around a Redis cache-aside pattern for redirect performance, with Celery handling asynchronous analytics processing.
Engineering Focus
- Redis cache-aside pattern for redirect resolution — cache hit avoids the database entirely
- Celery workers process click events asynchronously without adding latency to the redirect path
- Sliding window rate limiting using Redis sorted sets to prevent API abuse
- Dockerized deployment with Nginx as reverse proxy and gunicorn as WSGI server
- Streamlit-based analytics dashboard consuming the REST API for click trend visualization
PlantPulse
AI Plant Disease Detection Application
Overview
A full-stack application that serves a fine-tuned ResNet18 model via a Flask API to classify plant diseases from uploaded images. Demonstrates ML model integration into a production-ready backend with a Next.js frontend.
Engineering Focus
- ResNet18 fine-tuned on plant disease dataset, served via Flask with JSON prediction responses
- Image preprocessing pipeline normalizing input before model inference
- Low-latency inference endpoint optimized for single-image classification requests
- Cross-domain architecture connecting ML model serving (Flask/Python) with a React frontend
Vegacodes
Talent Acquisition Platform
Overview
A web platform bridging businesses with technical talent, featuring real-time job listings, authentication, and a modern React-based UI. Built on Firebase for backend services with Next.js for the frontend layer.
Engineering Focus
- Server-side rendering with Next.js for fast initial page loads and SEO optimization
- Firebase Firestore for real-time data synchronization and Firebase Auth for identity management
- Accessible component system built on Radix UI primitives with custom Tailwind styling
- Serverless architecture with Firebase Functions handling backend logic
Technical
Architecture Stack.
Engineered around Python backend architectures, high-throughput database layers, intelligent vector retrieval, and automated delivery pipelines.
Core Backend & Runtime
Standard backend language for services, algorithms, and data engines.
Architected scalable multi-app services with clean repository patterns.
Decoupled task processing with worker pools and retry strategies.
High-performance asyncpg query pipelines with explicit transaction bounds.
Persistence & In-Memory Stores
Optimized complex query execution plans, indexes, and asyncpg pools.
Sub-millisecond read caching, session stores, and Celery broker queues.
High-dimensional embedding storage for semantic similarity lookups.
Structured data storage with normalized schemas and integrity constraints.
Intelligent Systems & Retrieval
Custom scoring profiles, field boosting, and semantic re-ranking.
Context-augmented generation pipelines with chunking and re-ranking.
Serving fine-tuned computer vision and recommendation models via APIs.
Infrastructure & CI/CD
Hermetic container definitions for development and production workloads.
Reverse proxying WSGI/ASGI servers with optimized buffer settings.
Automated regression testing, linting, packaging, and versioned releases.
Frontend & UI Engineering
Capable of building responsive, production-ready interfaces to complement backend architectures.
Approach
Engineering Philosophy
How I think about building software systems.
Understand the problem.
Before writing a single line of code, understand the business constraints, scale requirements, and edge cases.
Design for maintainability.
Code is read far more often than it is written. Clear architecture and clean code over clever hacks.
Optimize only when necessary.
Avoid premature optimization. Identify true bottlenecks through profiling and metrics.
Measure before scaling.
Scale based on actual data, telemetry, and monitoring, not assumptions.
Prefer simplicity over cleverness.
The best systems are easily understood by the next engineer who has to maintain them.
Current Focus
Research & Architecture Exploration
Actively architecting systems and exploring high-leverage paradigms across modern backend infrastructure, distributed state management, and intelligent service integration:
- Distributed System Design & Resilience
- Developer Tooling & Automation Engines
- AI-Powered Application Architecture & RAG
- Cloud-Native Service Patterns & Async I/O
- API Design & Query Performance Optimization
Interested in backend engineering, search systems, or AI applications?
Let's connect.
Open for backend engineering opportunities, architectural discussions, and technical collaborations on search infrastructure, distributed services, and intelligent platforms.
[email protected]Gaurav Verma is an Application Programmer @ TCS and Python Backend Engineer specializing in Search Systems, Information Retrieval, Azure AI Search, Django, and Scalable Applications.
© 2026 Gaurav Verma. All rights reserved.