Software Engineer at Paytm

Abhay Rawat

Backend Engineer. AI & Agentic Systems. Software Engineer @ Paytm. Microservices at Scale

I build backend systems for high-traffic consumer fintech — and the agentic tooling that ships them faster.

See my work Resume Get in touch

Noida, India

35%
faster delivery
Agentic SDLC workflow on Claude + Cursor with Jira/Confluence MCP
20%
faster API response
Pricing microservice migration across Paytm Travel
10%
lower deploy cost
Infrastructure reduction on the same migration
1000+
DSA problems
Solved across LeetCode and GeeksforGeeks

Experience

Where I've shipped

Backend services in production at consumer scale, plus the tooling built around getting them there.

Software Engineer at Paytm

Jun 2025 — PresentNoida, India

End-to-end owner of Cancellation Protect and Travel Insurance within Paytm Travel — architecture, development, and production support on consumer-scale traffic.

35%faster feature delivery
Built an AI-powered agentic SDLC workflow on Claude and Cursor, wired to Jira and Confluence MCP servers, which also generates automated QA handoffs.
20%faster API response
Migrated the pricing microservice behind Cancellation Protect and Travel Insurance across Paytm Travel.
10%lower deployment cost
Cut infrastructure spend on the migrated pricing service as part of the same effort.
Architect, develop and maintain scalable backend services in Node.js, Express and Java Spring Boot for high-traffic consumer flows.

Node.js
Express.js
Java
Spring Boot
Microservices
Docker
Redis
AI Agents

Software Engineer Intern at Paytm

Jan 2025 — Jun 2025Noida, India

Backend development on financial transaction flows, with a focus on rule-driven decisioning and production observability.

Developed and optimized backend services in Node.js and Express, tuned for throughput and scalability.
Worked with the Drools rule engine to streamline dynamic decision-making in financial transaction paths.
Designed and deployed automated alerting for API and service failures, improving reliability and production monitoring.

Node.js
Express.js
Drools
SQL
Docker

Software Developer Intern at MentorAide

Jan 2024 — Jul 2024Remote

Full-stack product work alongside quantitative forecasting research for equity price prediction.

11.45%simulated return (LSTM)
Built LSTM and ARIMA forecasting pipelines for equity price prediction; the LSTM strategy outperformed ARIMA's 7.45% in simulated portfolio returns.
Led development and maintenance of scalable web applications in React and Redux, focused on robust interfaces and render performance.

React
Redux
Python
TensorFlow

Selected work

Things I've built

Side projects, mostly at the intersection of retrieval, agents, and backend plumbing.

NeuroHub

2024

LLM-powered learning management system

A full LMS built with a team of five, combining AI-generated-assignment detection with live face detection to keep assessments honest.

98% accuracy detecting AI-generated assignment submissions
30% increase in user engagement after the interface rework
Microservice split across auth, content, and inference

React
Node.js
Microservices
JWT
Hugging Face
TensorFlow
Python
AWS

Talk With Mongo

2024

Natural language to MongoDB queries

Translates plain English into NoSQL queries via a Hugging Face LLM and LangChain, then connects to a live database and runs them — schema introspection included.

Schema-aware prompting so generated queries match the live collection shape
Interactive interface for connecting and querying without writing NoSQL

Python
LangChain
Hugging Face
MongoDB

Document Q&A with RAPTOR

2024

Hierarchical RAG over textbook corpora

Extracts content from textbooks, builds a MILVUS vector store using RAPTOR hierarchical indexing, and answers questions over it with an LLM.

RAPTOR recursive summarization tree for multi-level retrieval
Handles questions that span chapters, not just single passages

Python
LangChain
MILVUS
RAG
Hugging Face

Also worth a look

AI-Generated Text Detector

2024

Transformer classifier, deployed on EC2

Distinguishes AI-generated from human-written text at 98.8% accuracy using a Hugging Face transformer architecture, served from AWS EC2.

Python
TensorFlow
Hugging Face
AWS

YouTube Video Assistant

2024

Context-aware Q&A over video transcripts

Takes a YouTube link, analyses the transcript, and answers arbitrary questions about the content with citations back to the source.

Python
LangChain
RAG

LSTM Equity Forecasting

2024

Sequence model with a backtested strategy

LSTM model forecasting equity prices, paired with a simple trading strategy that returned 11.45% in simulation.

Python
TensorFlow
pandas
yfinance

ARIMA Trading Strategy

2024

Classical baseline for the LSTM work

Statistical price prediction with ARIMA and a matching trading strategy — the 7.45% baseline the LSTM model was measured against.

Python
statsmodels
pandas

All repositories on GitHub

Toolkit

What I work with

Grouped by what they're for, rather than as an undifferentiated wall of logos.

Languages

JavaScript
Java
C++
SQL
TypeScript
Python

Backend & Frameworks

Node.js
Express.js
Spring Boot
REST APIs— Design, versioning, contract testing
React

Databases

MongoDB
Redis
MySQL
Cassandra— Wide-column store for high-write workloads

Architecture

Microservices— Service decomposition, inter-service contracts
System Design
Rule Engines— Drools — dynamic decisioning in transaction flows
Drools
MVC

DevOps & Tools

Docker
Git
Claude Code— Agentic SDLC workflows with MCP servers
Cursor

AI / ML & Automation

LangChain
LangGraph— Stateful multi-agent graphs
AI Agents— Tool-use loops, MCP server integration
RAG— Retrieval-augmented generation, RAPTOR indexing
Hugging Face
TensorFlow
NLP

Background

About

Software Engineer at Paytm, where I own Cancellation Protect and Travel Insurance end to end — architecture through production support. I work in Node.js, Express and Spring Boot on services that carry real consumer traffic, and I build agentic LLM tooling around the SDLC itself. Previously: LSTM forecasting pipelines, and an LMS with AI-generated-content detection.

Education

National Institute of Technology, Kurukshetra
B.Tech, Information Technology
CGPA 8.62 / 102021 — 2025
Maharishi Vidya Mandir, Sitapur
CBSE, Class XII (PCM)
95% aggregate2019 — 2020

Achievements

Certificate of Recognition, Paytm2025
Awarded for exceptional performance and contribution.
1000+ DSA problems solved
Across LeetCode and GeeksforGeeks.
Top performer — CSI RAIT Open Code2022
Open-source contribution program.
Smart India Hackathon2023
Selected at the college level.
Hackshetra finalist
24-hour hackathon.

Contact

Let's talk

Open to interesting backend and applied-AI problems. Happy to talk shop either way.