PBPriyabrata BeraStaff Frontend Engineer | Frontend Architect | Full-Stack Developer | AI Engineer
Immediate Joiner

Projects

GenAI applications, publisher tools, and large-scale frontend migrations

Selected work across GenAI apps, business admin tools, enterprise platforms, and Next.js migrations.

Projects

Featured product work and migrations

Featured

AI Conversational Assistant

End-to-end AI conversational app using Next.js/React.js and locally hosted Ollama LLMs, with streaming responses and conversation memory.

Next.jsFastAPIOllamaLLM
My role
Personal / Self-Driven
Key impact
Real-time streaming chat with an LLM provider abstraction layer and configurable system prompts.
Featured

RAG-Based Document Intelligence

Retrieval-Augmented Generation app for question answering over PDFs, with embeddings, vector search, and grounded, cited answers.

Next.jsPythonChromaDBRAG
My role
Personal / Self-Driven
Key impact
Grounded responses with source citations (document, page, snippet) and reduced hallucinations.
Featured

AI Personalized News Intelligence Platform

GenAI platform delivering personalized, topic-aware news feeds with embeddings, semantic search, and LLM-powered article summaries.

Next.jsFastAPIEmbeddingsOllama
My role
Personal / Self-Driven
Key impact
Personalized recommendations and concise summaries with private, local LLM inference.
Featured

Centralized Configuration Management Platform

Configuration management platform standardizing implementation patterns across 8 repositories at The Economic Times.

Next.jsReact.jsNode.js
My role
Tech Lead
Key impact
Cut integration time for new properties and standardized patterns across flagship products.
Project

Internal Admin Panels

Management and subscription admin panels giving business teams self-serve control over configurable content and data.

Next.jsReact.jsNode.js
My role
Tech Lead
Key impact
Removed engineering dependency for day-to-day content and subscription changes.
Project

XSLT to Next.js Migration

Large-scale migration of legacy XSLT applications to a modern Next.js architecture across flagship Economic Times products.

Next.jsSSRMigration
My role
Tech Lead
Key impact
Improved accessibility, maintainability, and Core Web Vitals scores.
Project

FedEx Print Online

Enterprise document printing, customization, and order management platform built at Infogain.

React.jsJavaScriptCanJSAJAX
My role
Sr. Consultant
Key impact
Delivered scalable printing and order management capabilities for a global enterprise customer.

Case Studies

Selected project case studies

RAG-Based Document Intelligence Application

Screenshots placeholder

Problem

Extracting reliable answers from large PDF documents is slow and error-prone, and general LLMs hallucinate without grounding in the source material.

Solution

Built an end-to-end RAG pipeline — PDF extraction, text chunking, embeddings, and ChromaDB vector storage — with semantic retrieval and grounded prompting, served through FastAPI and Ollama.

Tech Stack

Next.jsPythonFastAPIChromaDBOllamaRAG

My Role

Personal / Self-Driven

Key Features

  • PDF extraction and text chunking
  • Embedding generation with ChromaDB vector search
  • Grounded prompting to reduce hallucinations
  • Source citations with document, page, and snippet

Impact

Accurate, cited answers grounded in document context, with local streaming inference via Ollama and FastAPI.

AI Conversational Assistant

Screenshots placeholder

Problem

Building a real-time conversational AI that maintains context across turns without being locked to a single LLM provider.

Solution

Built a streaming chat application on Next.js/React.js with a FastAPI/Node.js backend, conversation memory, and an LLM provider abstraction layer.

Tech Stack

Next.jsReact.jsFastAPINode.jsOllamaLLM

My Role

Personal / Self-Driven

Key Features

  • Streaming AI responses
  • Conversation memory across interactions
  • LLM provider abstraction layer
  • Configurable system-prompt management

Impact

A responsive, provider-agnostic conversational experience running on locally hosted Ollama models.

Centralized Configuration Management Platform

Screenshots placeholder

Problem

Implementation patterns were inconsistent across properties and slow to extend to new repositories, lacking a single standard.

Solution

Designed and built a centralized configuration management platform (Next.js, React.js, Node.js) deployed across 8 repositories to standardize implementation.

Tech Stack

Next.jsReact.jsNode.js

My Role

Tech Lead

Key Features

  • Centralized configuration management
  • Cross-repository standardization
  • Self-serve business tooling
  • Reusable implementation patterns

Impact

Standardized implementation across 8 repositories and cut integration time for new properties.

Legacy XSLT to Next.js Migration

Screenshots placeholder

Problem

Flagship pages ran on legacy XSLT, limiting accessibility, maintainability, and further performance improvements.

Solution

Directed a large-scale migration to React.js, Next.js, and TypeScript with a focus on SSR and Core Web Vitals.

Tech Stack

Next.jsReact.jsTypeScriptSSR

My Role

Tech Lead

Key Features

  • SSR-based rendering
  • Improved Core Web Vitals
  • Modern component architecture
  • Accessibility improvements

Impact

Measurable Core Web Vitals gains and a more maintainable codebase across flagship products.