Anupam
Kumar

Backend & Generative AI Engineer

Python · AWS · Azure · Kubernetes

Backend engineer who builds Generative AI systems and ships them to production on AWS and Kubernetes.

Work hardThink harderPlan hardest

About

The engineer behind the systems.

I'm a backend engineer who's spent the last couple of years building Generative AI systems and shipping them to production. Not prototypes — real products with real users. Two GenAI products live at Q3 Technologies, a multi-agent chatbot delivered for a client, and a 14-stage document-processing pipeline running on Azure Kubernetes Service.

Most of my work lives at the application layer of LLMs: RAG pipelines, multi-agent orchestration, document AI, and the retrieval and reliability work that keeps them dependable. I'm equally comfortable in the infrastructure underneath — serverless on AWS, AKS and KEDA on Azure, and the CI/CD that keeps deploys boring.

NIT Hamirpur CSE graduate. I like hard system-design problems with real deadlines.

Location

Gurugram, India

Currently

SDE @ Q3 Technologies

Oct 2023 – Present

Education

B.Tech CSE — NIT Hamirpur

2019 – 2023 · CGPA 8.41/10

Achievements

900+ DSA problems

CF rank 280 · CodeChef rank 121 · JEE 99.08%ile

2 Production GenAI products

~1,000 Daily users at 99.5% uptime

124+ Lambda functions auto-deployed

~$9K/yr AWS cost savings

14-stage Multi-tenant AI pipeline

900+ DSA problems solved

Skills

The stack I reach for

Grouped by how deep I actually go, not just what I've touched.

Generative AI & LLMs

Core · what I build daily

LangChainRAGMulti-Agent SystemsAWS BedrockAzure OpenAIAzure AI FoundryPrompt EngineeringFunction CallingEmbeddingsVector SearchStructured OutputsMCP

Languages

Python core · others as needed

PythonTypeScriptJavaScriptSQLBashC++

Backend & APIs

Core · what I build daily

FastAPIDjangoFlaskREST APIsGraphQLWebSocketServer-Sent Eventsasync/awaitPydantic v2OpenAPI

DevOps & Platform

Production · shipped & owned

DockerKubernetesCI/CDGitHub ActionsBitbucket PipelinesCloudFormationKustomizepytestmypySonarQube

AWS

Production · shipped & owned

LambdaAPI GatewayAppSyncDynamoDBS3SQSSNSEventBridgeCognitoBedrockAuroraServerless Framework

Microsoft Azure

Production · shipped & owned

AKSKEDAAzure AI SearchAzure OpenAIDocument IntelligenceAI FoundryService BusEvent GridBlob StorageManaged Identity

Databases

Production · shipped & owned

PostgreSQLDynamoDBQdrant (vector)Azure AI Search (vector)RedisAurora PostgreSQL

Projects

Things I've shipped

Production GenAI and backend systems — live products with real users, not prototypes.

Australia Feb 2026 – Present

AI Bid-Writing Platform

14-stage multi-tenant document pipeline; 3 production microservices on Azure Kubernetes Service with KEDA autoscaling.

More detail

Owned the indexer, extraction, and summarisation consumers of a 14-stage multi-tenant pipeline on AKS — live in two months. Eliminated Azure OpenAI 429s under peak load with token-aware embedding batching (100 texts / 15k tokens) and Retry-After backoff. Added a stale-job reaper, graceful SIGTERM handling, and a multi-tenant cache on Pydantic v2 and asyncpg. Trimmed the Docker image 23% (775→600 MB) and added a DOCX→PDF→Document Intelligence fallback via headless LibreOffice.

AI Bid-Writing Platform architecture diagram

PythonFastAPIAzure AKSKEDAAzure Service BusAzure AI SearchAzure OpenAIDocument Intelligence +3

Internal Project Jan 2026 – Feb 2026

Internal Contracts AI

Production RAG backend for contracts Q&A — JWT-secured, SSE streaming, multi-page citations. Live and used daily internally.

More detail

Took the backend from an empty repo to production in five weeks: a JWT-secured RAG pipeline with SSE streaming, multi-page citations, async attachment indexing, and checksum dedup on uploads. Live and used daily by senior internal teams. Security hardened on the way: pulled an SSO bypass, added CORS, rate limiting, security headers, and intent-based query routing.

FastAPIPostgreSQLQdrantJWTLangChainAWS BedrockOllama

EdTech Client Dec 2025 – Jan 2026

Multi-Agent LMS Assistant

Multi-agent Canvas LMS assistant — 4-agent system (intent classifier, internet-fallback, document, identity) built on Azure AI Foundry.

More detail

Delivered the first release; the client came back with a follow-on proposal and a bigger scope. Designed the 4-agent setup for Canvas LMS Q&A. Refactored the intentAgent to clear every SonarQube issue across 7+ PRs, and kept the agents model-independent so swapping LLM versions doesn't break things.

PythonFastAPIAzure AI FoundryLangChainCanvas LMS API

UK Rail Operator Jul 2025 – Nov 2025

Real-Time Train Tracking

Live-tracking via AppSync subscriptions over DynamoDB streams. Serves real-time berth updates across regional rail.

More detail

Cut DynamoDB network overhead 25% and Lambda runtime 20% by reworking batch operations, and rebuilt the SVG upload path with S3 presigned URLs to get past the 10 MB cap. Built the live-tracking layer on AppSync subscriptions over DynamoDB streams. Cleared three production incidents.

PythonAWS LambdaDynamoDBAppSyncAPI GatewayS3

UK rail operator Dec 2023 – mid-2025

UK Rail Delay-Repay Claims Portals

~1,000 daily users, 99.5% uptime. Claims portals (customer + agent workflows) on AWS Lambda for a UK rail operator.

More detail

Built the delay-repay claims portals on AWS Lambda. Wired in Loqate for UK address and IBAN validation, shipped 38+ Lambda functions through CI/CD, and cut claim processing time 40% with conditional CSV and PDF reports off Aurora Postgres.

PythonAWS LambdaServerless FrameworkCloudFormationAurora PostgreSQLLoqate API

Open Source Personal project

HealAll India

Open-source platform helping people find medical assistance across India. Built with Next.js, React, and interactive maps.

More detail

Personal open-source project built to help people find medical assistance. Features interactive maps for locating nearby medical resources, Google OAuth, and real-time location data. Built with Next.js, React, Leaflet, and Zustand — deployed on Vercel.

Next.jsReactTypeScriptLeafletZustandVercel

Experience

Where I've worked

Software Development Engineer

Q3 Technologies (Q3 Infotech Pvt Ltd)

AI bid-writing (AU) · Internal Contracts AI (RAG chatbot) · Multi-agent LMS Assistant · Real-time train tracking & UK delay-repay claims

Oct 2023 – Present Gurugram, India

Shipped 3 production microservices on Azure Kubernetes Service with KEDA autoscaling, plus a 14-stage multi-tenant document pipeline — live in two months.
Own the backend and architecture across two production GenAI products and a multi-agent client build, spanning RAG, vector search, and document AI.
Cut AWS iteration cycles in half and saved ~$9,000/year with a Docker-based Lambda debugger. Took manual Lambda provisioning down ~75% with a CI/CD pipeline auto-deploying 124+ functions.
Built `setuptemplate`, a pip-installable scaffolder for FastAPI/Flask/Django with mypy, pytest, and flake8 — turns new-project setup from hours into under two minutes.
Fixed three critical production incidents and wrote post-mortems the team still leans on. Mentor three junior engineers on AWS and the GenAI stack.

Research Intern

Solid State Physics Laboratory (SSPL), DRDO

Jun 2022 – Aug 2022 New Delhi, India

Studied GaN HEMT performance for telecom use with Silvaco simulations and TonyPlot, and co-authored a comparison against silicon benchmarks that fed into device-design decisions at SSPL.

Writing