/ Software Engineer
Software engineer
at IBM.
I work across backend, cloud and applied AI, building serverless systems that bring large language models into production.
The kind of serverless GenAI pipeline I build at IBM, on AWS Lambda, Amazon Bedrock, DynamoDB and S3.
/ Selected work
Systems I build at IBM.
Production GenAI on a serverless AWS stack: retrieval, guardrails, model tuning, and the delivery pipeline behind them.
Generative AI integration
Integrating large language models (Amazon Bedrock and Anthropic Claude) into enterprise applications, with multiple retrieval strategies and an extended-reasoning mode.
Retrieval-augmented generation
RAG over domain knowledge bases, embeddings and semantic retrieval with content segregation by business function, for grounded, relevant answers.
AI guardrails & validation
Automated validation with feedback loops keeps model outputs accurate, safe and on-policy, turning probabilistic models into dependable workflows.
Model tuning & prompt engineering
Migrating and tuning models with dynamic prompt configurations and evaluation, getting the right behaviour from the right model for each market.
Serverless AWS backend
Python on AWS Lambda with S3, DynamoDB, SQS and API Gateway. Scalable, event-driven services powering the AI features.
CI/CD & delivery
Shipping through a disciplined pipeline (Jenkins, SonarQube and JFrog) with Git branching across development, QA and production.
/ Experience
A short history.
Software Engineer Full-time
Integrate LLMs (Bedrock, Claude) into a production GenAI platform: RAG, guardrails and model tuning. Build the serverless AWS backend and ship via CI/CD with disciplined Git branching.
Cloud Full Stack Apprenticeship
Cloud full-stack development across the same platform, ramping into serverless architecture, AWS services and disciplined Git workflows.
Full Stack Developer TeleMatrix Internship
Built web applications end to end: front end in HTML/CSS, JavaScript and React, back end in Node.js, Express and Python.
/ Capabilities
What I work with.
AI / GenAI
- Amazon Bedrock
- Anthropic Claude
- RAG & knowledge bases
- Guardrails & validation
- Prompt engineering & tuning
- Embeddings (Titan, Cohere)
Cloud / Backend
- Python
- AWS Lambda, Serverless
- S3, DynamoDB, SQS
- SES, Cognito, API Gateway
- Textract, REST APIs
- Microservices
DevOps / CI-CD
- Jenkins, SonarQube, JFrog
- Git, Bitbucket
- Docker, ECS
- CloudWatch
Frontend
- JavaScript (ES6+)
- React.js
- HTML & CSS
- Responsive design
Databases
- DynamoDB, NoSQL modeling
- MySQL
Languages
- Python
- JavaScript
- C++
- SQL, Java
/ About
The short version.
I'm a software engineer at IBM, where I integrate large language models into a production GenAI platform on AWS. I build serverless services with Lambda, DynamoDB, SQS and Amazon Bedrock, shipping retrieval-augmented generation, guardrails and model tuning into real enterprise workflows.
A big part of the job is tracing failures across services and fixing them at the root: debugging model responses, resolving data inconsistencies, and hardening reliability. Before IBM I was a full-stack developer, comfortable from the React front end down to the queue.
- Education
- B.Tech, ECE · Heritage Institute of Technology · CGPA 8.6
- Languages
- English · Hindi · Urdu · Bengali
- Based in
- Kolkata, India