I am a Senior Software Engineer with 7+ years of experience building and shipping production applications across banking, media, education, e-commerce, and consumer products.

My core engineering expertise is in React Native, React, TypeScript, JavaScript, application architecture, performance optimization, API integration, and cross-platform application development.

Alongside my software engineering background, I have developed hands-on expertise in Generative AI, Agentic AI, and backend development, working with:

• Retrieval-Augmented Generation (RAG)
• LLM applications and prompt engineering
• AI agents and tool calling
• LangChain and LangGraph
• Model Context Protocol (MCP)
• Embeddings and vector search
• Hybrid retrieval and reranking
• Structured outputs with Pydantic
• Memory and state management
• LLM evaluation
• Python and FastAPI
• AWS and Docker

I enjoy building AI systems that combine LLM reasoning with deterministic tools, backend services, retrieval pipelines, and real user interfaces.

A voice-based AI assistant designed for conversational healthcare interactions.

• Real-time Speech-to-Text → LLM/RAG → Text-to-Speech pipeline
• Conversational voice interaction
• Intent understanding and tool execution
• RAG-based contextual responses
• Conversation context and state management
• Backend orchestration using FastAPI
• Designed for integration with a mobile voice experience
• Structured AI responses and backend tool execution

Tech: Python · FastAPI · LangChain · RAG · LLMs · Embeddings · Vector Database · STT · TTS · WebSocket

An AI-powered platform for managing civic complaints such as potholes, garbage, water leakage, and broken infrastructure.

• AI-powered citizen and admin workflows
• Custom MCP server and MCP client
• Complaint intake and categorization
• Duplicate complaint detection
• SLA-based prioritization
• Worker assignment
• Notifications and administrative workflows
• AI-powered admin summaries
• Tool-driven agent workflows
• Separation of LLM reasoning from deterministic business operations

Tech: TypeScript · Node.js · MCP SDK · Gemini · REST APIs · Supabase · AI Agents · Tool Calling

An AI-powered experience that allows users to ask natural-language questions about book content and receive grounded answers.

• Document ingestion pipeline
• Hierarchical document chunking
• Embedding generation
• Vector search
• Metadata-aware retrieval
• Retrieval and reranking
• Context-aware prompt construction
• Grounded LLM responses with source references
• Retrieval and response evaluation
• AI recommendation capabilities based on reading and content signals

Tech: React Native · TypeScript · Python · FastAPI · RAG · Embeddings · Vector Search · LLMs · AWS · Docker

A conversational property-discovery assistant that allows users to search properties and perform actions through WhatsApp.

• WhatsApp webhook integration
• Natural-language property search
• Gemini-powered intent understanding
• Tool calling for structured property operations
• Property search and filtering
• Property photo delivery through WhatsApp
• Session and conversation state
• Lead capture
• Site-visit booking
• Admin dashboard

Tech: Python · FastAPI · Gemini · Meta WhatsApp Cloud API · SQLite · AI Agents · Tool Calling

An AI learning assistant built around course material with a focus on retrieval quality and evaluation.

• Document ingestion and parsing
• Recursive and hierarchical chunking
• Dense embeddings
• Vector search
• Sparse retrieval using TF-IDF / BM25-style techniques
• Hybrid retrieval
• Cross-encoder reranking
• Structured outputs
• Retrieval evaluation
• Groundedness and relevance evaluation
• Base-model vs fine-tuned-model comparison

Evaluated retrieval quality across test questions using metrics such as:

• Chapter Hit@5
• Source Hit@5
• Retrieval relevance
• Groundedness
• Completeness
• Hallucination detection

Tech: Python · LangChain · LangGraph · Pydantic · Embeddings · Vector Databases · Hybrid Retrieval · Reranking · RAG · LLM Evaluation

An AI-assisted insurance claim processing backend focused on document ingestion, OCR, parsing, chunking, and structured claim data.

• Claim and policy domain models
• Document upload and validation
• Document processing lifecycle
• PDF and DOCX parsing
• Image processing
• OCR fallback using Tesseract
• Document page extraction
• Recursive chunking
• Section-aware chunking
• Claim document metadata
• Processing and debugging APIs
• SQLite-backed persistence
• Automated test coverage

Tech: Python · FastAPI · PyMuPDF · Python-docx · Pillow · Tesseract OCR · SQLite · Pydantic

A GenAI-powered system for SMEs combining financial, inventory, sales, and external intelligence to support business decisions.

• Demand prediction workflows
• Inventory intelligence
• Financial analysis
• Weather-aware demand analysis
• RAG-style retrieval
• Vector search
• Data processing pipelines
• Multi-agent architecture
• Decision-oriented AI workflows

Data Agent
    ↓
Weather Agent
    ↓
Decision Agent
    ↓
Business Recommendation