Consulting data platform
Nagarro · Consulting · AWS, Terraform, LangChain, Kafka, Snowflake
Delivered a consulting engagement spanning Terraform-managed AWS infrastructure, a production LangChain RAG bot, realtime Kafka streaming into Snowflake, and agent skills that automate engineering tasks.
Highlights
- Provisioned and maintained AWS workloads with Terraform (Lambda, S3, API Gateway).
- Deployed a RAG chatbot on Snowflake Cortex AI with LangChain parsing into Snowflake vector tables.
- Built Kafka event flows with Lambda processing into Snowflake for realtime analytics.
- Created agent skills and automations to speed up delivery, reviews, and ops.
Back to portfolio
Azure & Databricks modernization
Nagarro · Insurance · Azure Data Factory, Databricks, PySpark
Led modernization of legacy on-premises pipelines onto Azure for a major insurance program: ADF orchestration, Databricks ETL, and reliable batch ETL/ELT at scale.
Highlights
- Migrated 1+ TB to Azure SQL and Snowflake with zero downtime; led a team of 3.
- Implemented Azure Data Factory orchestration and Databricks / PySpark transformation pipelines.
- Enforced data governance, quality SLAs, and observability on production Azure workloads.
Back to portfolio
Serverless media migration
Accenture · Media & Entertainment · Lambda, SQS, DynamoDB, S3
Designed an economical AWS migration architecture for ingesting and processing on-prem media at scale using serverless, fault-tolerant workflows.
Highlights
- Event-driven pipelines with Lambda, SQS, and DynamoDB.
- Automated framework to transfer 1.5 million media assets safely.
- Compression/upload utility for 60GB packages into the DAM system.
- Kept migration overhead near $100 by maximizing serverless and minimizing EC2.
Back to portfolio
AWS data lake & reporting
Deloitte · Nutrition · Glue, Athena, S3, SageMaker
Delivered big-data solutions tied to business needs: data quality fixes, optimized processing, and reliable lakehouse patterns on AWS.
Highlights
- ETL on the data lake with Glue, crawlers, and Athena.
- Hadoop sandboxes with Glue Dev Endpoints and SageMaker notebooks.
- Pipelines routing data into multiple ML models.
- Near real-time reporting for executive decision-making.
Back to portfolio
AI Travel Planning Assistant
Python · LangChain · Azure OpenAI · ChromaDB · Streamlit · Docker
Context-aware AI assistant for trip planning (Singapore focus), combining a document knowledge base with real-time data via MCP tools.
Highlights
- TravelAgent orchestrator with intent classification, multi-turn context, and dual paths: RAG (ChromaDB) and MCP (weather, currency).
- Streamlit UI; responses attribute knowledge-base facts, live MCP data, and LLM synthesis.
- Azure OpenAI LLM, local HuggingFace embeddings, Docker Compose deployment; pytest test suite.
View Source Code
Back to portfolio
Football Match Data Scraper & ETL
Python · Beautiful Soup · Pandas · Web Scraping
Automated web scraping and transformation pipeline to retrieve football match statistics and export clean CSV for analysis.
Architecture
- Extraction: Beautiful Soup scraper for dynamic match pages.
- Transform: Pandas normalization of team and player metrics.
- Load: Standardized CSV for downstream tools.
View Source Code
COVID-19 India Data Analysis
Python · Pandas · API Integration · Matplotlib
End-to-end flow from a public REST API to processed time series and an animated racing bar chart.
Pipeline
- Reliable JSON ingestion from a live API.
- Pandas processing with rolling averages and anomaly handling.
- Matplotlib animation for the final visualization.
View Source Code
Event management platform
Personal · AI chat · Attendance · Networking
Built a scalable event management system where attendees use AI chatbots, create connections with other participants, mark attendance, and network during events.
Highlights
- AI chatbots for attendee support and event information.
- User connections and networking features across events.
- Attendance tracking integrated into the event flow.
- Architecture designed for growth, built with AI-assisted development for rapid iteration.
Back to portfolio
Fruit Classifier | Transfer Learning
Python · Keras · AWS · Deep Learning
MTech thesis project using pre-trained CNN architectures on AWS for image classification.
Approach
- Transfer learning with Keras (ResNet/VGG-style backbones).
- Preprocessing and training on AWS GPU compute.
- Fine-tuned classification head and hyperparameters.
View Source Code