Manav Arora profile photo 1 on Tal

Manav Arora

Backend Engineer | Distributed Systems • Microservices | High-Scale API Optimization | Multi-Region E-Commerce Integrations | AWS/Azure | AI/ML & RAG

Delhi

6+ years of work experience

HashedIn by Deloitte
Aug 20242 yrs 1 moPresent

Bengaluru, Karnataka, India

Architected horizontally-scalable WebSocket proxy using Python, FastAPI, and Redis pub/sub fanout handling 8K+ concurrent connections at 50K msg/sec with P95 latency under 50ms and 87% cache hit ratio. Implemented cloud- agnostic factory pattern eliminating vendor lock-in, saving $50K+ in Azure costs. Deployed across 7+ teams with HPA-based auto-scaling at 99.8% uptime. • Resolved critical production Out-of-Memory crashes via heap dump analysis and memory profiling — traced root cause to unnecessary deep object copying in processing pipeline. Implemented object pooling and reference-based handling, achieving 70% memory footprint reduction and eliminating all OOM incidents across 5 dependent teams. • Designed Backend-for-Frontend (BFF) system integrating WebSockets for real-time updates and Kafka for reliable async message passing between microservices. Reduced frontend integration complexity by 60% with a single consistent API layer enabling seamless multi-service scaling. • Built high-throughput data simulator in Java with JEXL templating processing 250K+ records in under 45s (5.5K rec/sec) with dual-mode architecture: streaming with backpressure and batch with JSON serialization. Eliminated critical SAP/Highbyte dependencies for 5+ teams. • Integrated NVIDIA Cosmos 7B and VLM models for real-time video analysis achieving 92% detection accuracy at 18 FPS with P95 inference latency of 210ms. Implemented severity-based event classification with automated alerting. • Lead architecture reviews, technical design decisions, and conduct technical interviews. Participated in AWS to Azure migration of a Node.js microservices platform, gaining hands-on depth across both cloud ecosystems. Recognised with 4 performance awards in 18 months for the data simulator & OOM resolution, WebSocket proxy, and code quality contributions.
Anycover
Jun 20222 yrs 2 mosJul 2024

Remote

Optimized warranty widget API serving 2M+ daily requests (peak: 100 req/sec) across 50+ stores and 9 regions. Reduced P95 latency from 4s to 700ms (82% improvement) via Cache-Aside pattern with Redis (75% hit ratio), PostgreSQL query batching (2.5s → 300ms), and async S3 processing. Achieved 60% database load reduction at 99.2% uptime. • Built multi-region transaction platform supporting 9 currencies across 3 timezones with ACID-compliant transactions and real-time forex integration (P95 under 500ms). Handled 50K daily transactions (peak: 200/min) with zero failures through location-based routing and idempotent transaction design. • Engineered 6 production-grade e-commerce connectors in Node.js (Shopee, Lazada, Shopify, WooCommerce, Magento, Wix) with hybrid integration: webhooks for real-time sync (under 2min lag), polling with exponential backoff (95% retry success), and SQS queues for offline merchants. Averaged 1.5 weeks per integration. • Deployed production RAG system using FastAPI with pgvector and sentence-transformers (384-dim, cosine similarity 0.7 threshold) achieving 78% precision@5 and P95 latency under 800ms, reducing manual support workload by 40%. • Established CI/CD pipelines with GitHub Actions for multi-region AWS deployments with Datadog APM monitoring. Built automation suite including OCR invoice extraction and reconciliation dashboards, reducing financial close from 15 days to 3 days.
Anycover
Jun 202211 mosApr 2023

HelloSivi Software Labs
Jan 20222 mosFeb 2022

Remote

Enhancing System Performance and Scalability: Explored and implemented GraphQL systems, deploying backend on AWS to enhance performance and scalability by 30%. Gained proficiency in containerization with Docker, improving development workflows and system reliability.
Kredo Analytics Private Limited
Oct 20217 mosApr 2022

Remote

• Full Stack Development and Deployment: Worked on User Management System (UMS) and created Purchase Operations Sales Operations (POSO) using Python FastAPI and ReactJS, implemented role-based authorization for security, and leveraged Docker for streamlined deployment and improved workflow.
The Placement Cell, Keshav Mahavidyalaya
Jun 20211 yr 1 moJun 2022

India

The Placement Cell, Keshav Mahavidyalaya
Jan 20201 yr 6 mosJun 2021

New Delhi Area, India

GitHub
@manav-1

Manav’s AI Stack

antigravity
antigravity
cursor
cursor

Education

Keshav Mahavidyalaya - Delhi University

Bachelor of Science - BS, Computer Science

Keshav Mahavidyalaya - Delhi University

20193 yrs2022
Shaheed sukhdev college of business studies

Certificate, Data Analysis for Business

Shaheed sukhdev college of business studies

20192019

Join Manav and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Manav Arora on Tal

Headline

Backend Engineer | Distributed Systems • Microservices | High-Scale API Optimization | Multi-Region E-Commerce Integrations | AWS/Azure | AI/ML & RAG

About

Backend Engineer with 3+ years architecting distributed systems and real-time infrastructure. Built production platforms processing 250K+ records at 5.5K rec/sec, optimized high-traffic APIs serving 2M+ daily requests with 82% latency reduction, and designed multi-region transaction systems achieving zero downtime across 9 currencies.

Location

Delhi

AI stack

antigravity

cursor

Work experience

Software Engineer II at HashedIn by Deloitte, Aug 2024 - Present · 2 yrs 1 mo, Bengaluru, Karnataka, India, Architected horizontally-scalable WebSocket proxy using Python, FastAPI, and Redis pub/sub fanout handling 8K+ concurrent connections at 50K msg/sec with P95 latency under 50ms and 87% cache hit ratio. Implemented cloud- agnostic factory pattern eliminating vendor lock-in, saving $50K+ in Azure costs. Deployed across 7+ teams with HPA-based auto-scaling at 99.8% uptime. • Resolved critical production Out-of-Memory crashes via heap dump analysis and memory profiling — traced root cause to unnecessary deep object copying in processing pipeline. Implemented object pooling and reference-based handling, achieving 70% memory footprint reduction and eliminating all OOM incidents across 5 dependent teams. • Designed Backend-for-Frontend (BFF) system integrating WebSockets for real-time updates and Kafka for reliable async message passing between microservices. Reduced frontend integration complexity by 60% with a single consistent API layer enabling seamless multi-service scaling. • Built high-throughput data simulator in Java with JEXL templating processing 250K+ records in under 45s (5.5K rec/sec) with dual-mode architecture: streaming with backpressure and batch with JSON serialization. Eliminated critical SAP/Highbyte dependencies for 5+ teams. • Integrated NVIDIA Cosmos 7B and VLM models for real-time video analysis achieving 92% detection accuracy at 18 FPS with P95 inference latency of 210ms. Implemented severity-based event classification with automated alerting. • Lead architecture reviews, technical design decisions, and conduct technical interviews. Participated in AWS to Azure migration of a Node.js microservices platform, gaining hands-on depth across both cloud ecosystems. Recognised with 4 performance awards in 18 months for the data simulator & OOM resolution, WebSocket proxy, and code quality contributions.

Backend Engineer at Anycover, Jun 2022 - Jul 2024 · 2 yrs 2 mos, Remote, Optimized warranty widget API serving 2M+ daily requests (peak: 100 req/sec) across 50+ stores and 9 regions. Reduced P95 latency from 4s to 700ms (82% improvement) via Cache-Aside pattern with Redis (75% hit ratio), PostgreSQL query batching (2.5s → 300ms), and async S3 processing. Achieved 60% database load reduction at 99.2% uptime. • Built multi-region transaction platform supporting 9 currencies across 3 timezones with ACID-compliant transactions and real-time forex integration (P95 under 500ms). Handled 50K daily transactions (peak: 200/min) with zero failures through location-based routing and idempotent transaction design. • Engineered 6 production-grade e-commerce connectors in Node.js (Shopee, Lazada, Shopify, WooCommerce, Magento, Wix) with hybrid integration: webhooks for real-time sync (under 2min lag), polling with exponential backoff (95% retry success), and SQS queues for offline merchants. Averaged 1.5 weeks per integration. • Deployed production RAG system using FastAPI with pgvector and sentence-transformers (384-dim, cosine similarity 0.7 threshold) achieving 78% precision@5 and P95 latency under 800ms, reducing manual support workload by 40%. • Established CI/CD pipelines with GitHub Actions for multi-region AWS deployments with Datadog APM monitoring. Built automation suite including OCR invoice extraction and reconciliation dashboards, reducing financial close from 15 days to 3 days.

Jr Full Stack Developer at Anycover, Jun 2022 - Apr 2023 · 11 mos

Javascript Developer at HelloSivi Software Labs, Jan 2022 - Feb 2022 · 2 mos, Remote, Enhancing System Performance and Scalability: Explored and implemented GraphQL systems, deploying backend on AWS to enhance performance and scalability by 30%. Gained proficiency in containerization with Docker, improving development workflows and system reliability.

Full Stack Developer at Kredo Analytics Private Limited, Oct 2021 - Apr 2022 · 7 mos, Remote, • Full Stack Development and Deployment: Worked on User Management System (UMS) and created Purchase Operations Sales Operations (POSO) using Python FastAPI and ReactJS, implemented role-based authorization for security, and leveraged Docker for streamlined deployment and improved workflow.

Head Of Technical Operations at The Placement Cell, Keshav Mahavidyalaya, Jun 2021 - Jun 2022 · 1 yr 1 mo, India

Placement Coordinator at The Placement Cell, Keshav Mahavidyalaya, Jan 2020 - Jun 2021 · 1 yr 6 mos, New Delhi Area, India

Education

Bachelor of Science - BS, Computer Science, Keshav Mahavidyalaya - Delhi University

Certificate, Data Analysis for Business, Shaheed sukhdev college of business studies

Public links

github: https://github.com/manav-1