π London, UK Β |Β π§ arnabbiswas66@gmail.com Β |Β π linkedin.com/in/arnab-biswas-
I'm an Enterprise Solutions Architect with 20+ years of experience designing and delivering large-scale digital platforms in Publishing, eCommerce, and Cloud. I combine deep engineering expertise with strategic architecture thinking β supercharged by postgraduate qualifications in AI/ML (MSc, Liverpool John Moores University & PG Diploma, IIIT Bangalore).
Currently serving as eCommerce Domain Architect at John Wiley & Sons, leading headless commerce platforms, AI-powered recommendation engines, and cloud-native microservice ecosystems. My latest work explores Agentic AI β building and deploying custom agents on Amazon Bedrock AgentCore with FastAPI and LangChain.
- π Currently working on Agentic AI with Amazon Bedrock AgentCore + LangChain
- π± Exploring multimodal AI, attention mechanisms & LLM-powered agents
- ποΈ Architected 20+ enterprise platforms across eCommerce, CMS, IAM & Payments
- π€ Passionate about bridging ML/AI research with production-grade engineering
- β‘ Fun fact: I've built systems that serve millions of researchers and authors worldwide
- Supervised Learning: Linear/Logistic Regression, NaΓ―ve Bayes, SVM, Decision Trees, Random Forest, AdaBoost, GradientBoost, XGBoost
- Unsupervised Learning: Clustering, PCA, Dimensionality Reduction
- Deep Learning: ANN, CNN, RNN, LSTM, GRU β including 3D-CNN for video gesture recognition
- Multimodal AI: Text + image fusion with multi-level attention (MHA + Luong Attention) β see FND-MLA
- NLP: NLTK, SpaCy, Word2Vec, GloVe, FastText, Gensim
- Reinforcement Learning: Classical RL, Deep Q-Networks (DQN)
- Conversational AI: Rasa (chatbot development & deployment)
- Agentic AI: Amazon Bedrock AgentCore, LangChain, LangGraph, FastAPI agent servers
- Frameworks: TensorFlow, Keras, Scikit-learn, OpenCV, Pandas, NumPy
ML use-cases delivered: Lead Scoring Β· Customer Churn Β· Recommendation Systems Β· Customer Segmentation Β· Multimedia Spend Models Β· Next Best Offer Β· Cross-sell/Up-sell Β· Fake News Detection
- AWS β production-grade, PCI-compliant systems (EC2, ECR, Lambda, Kinesis, SQS, DynamoDB, Bedrock AgentCore)
- Serverless & Edge β AWS Lambda, Serverless Framework, Cloudflare Workers, Netlify Functions
- Amazon Bedrock AgentCore β custom agent runtimes, ARM64 Docker deployments, invocation pipelines
- Headless, decoupled microservice platforms at enterprise scale
- Custom Identity & Access Management (OAuth2, OpenID Connect, SAML, SSO)
- Cloud-native log management pipelines (ELK Stack + Kafka) processing millions of events
- Architect of headless commerce engines for B2C and B2B2C publishers
- Payment solutions integrating Stripe, Apple Pay, Google Pay, AliPay β fully PCI-compliant
- Product Information Management (PIM) and Product Lifecycle Management (PLM)
- MarTech integration, personalisation engines and recommendation systems
- Solution Architecture, Enterprise Architecture, Domain-Driven Design
- Microservices, Micro Frontends, API-first design (REST, GraphQL, SOAP), Event-driven systems
- Apache Kafka, JMS, Kinesis for high-throughput streaming
- Databases: MySQL, Oracle, ElasticSearch, MarkLogic, Neo4j, Redis
Languages
AI / ML / Deep Learning
Agentic AI & LLMs
NLP
Visualisation
Cloud & DevOps
Web Frameworks & APIs
Web & App Servers
Databases
Security
Streaming & Messaging
Methodologies
Collaboration & Tooling
| Project | Description | Tech |
|---|---|---|
| π€ AgentCore with FastAPI | Custom Amazon Bedrock AgentCore agent built with FastAPI + LangChain; ARM64 Docker container with /invocations & /ping endpoints for production deployment |
Python, FastAPI, LangChain, AWS Bedrock, Docker |
| π Multimodal Fake News Classifier | Novel FND-MLA model fusing text + images via Multi-Head Attention & Luong Attention for fake news detection (Twitter MediaEval & Weibo datasets) | TensorFlow, Keras, NLP, CNN, Attention Mechanisms |
| π¦ Continuous Delivery | CI/CD pipeline exploration | DevOps |
| π Netlify Feature Tour | Modern web deployment with Astro on Netlify | Astro, JavaScript |
| βοΈ JSON Rule Engine PoC | Proof-of-concept for a JSON-driven business rules engine | JavaScript |
Enterprise-scale projects (professional):
- π¦ PCI-Compliant Payment Platform β Stripe, Apple Pay, Google Pay, AliPay on AWS
- π Custom IAM + SSO Platform β OAuth2/OpenID Connect microservices at global scale
- π Central Log Intelligence β ELK + Kafka + Neo4j CMDB for outage prediction
- π Headless Commerce Engine β Decoupled B2C/B2B2C platform (SAP Hybris + cloud-native)
- π Scientific Research Platform β Serving millions of authors and researchers worldwide
| Qualification | Institution | Year |
|---|---|---|
| MSc, Artificial Intelligence & Machine Learning | Liverpool John Moores University | 2022β2023 |
| PG Diploma, ML & AI | IIIT Bangalore | 2020β2021 |
| BTech, Computer Science | Rajabazar Science College | 2001β2004 |
"Architecture is the art of making the complex feel simple β and AI is what makes that complexity worth solving."