Projects — Vittu Ramadasu Darshan

Automated Real Time Hospital Patient Data Analytics Platform on AWS

Developed automated ETL workflows within cloud-based internal systems using structured project management and stakeholder management supporting change management initiatives, delivering clear business value through faster analytics ensuring data management and accuracy.

AWS (S3, Lambda, Glue, QuickSight) Python SQL

GitHub Live Demo

NF Spec Manager - A Unified, Automated, and Governed Solution for Network Function Specification Management

Built a unified governed platform for managing Network Function specifications — featuring role-based access control, schema validation, automated dev→staging→prod promotion pipeline, audit logging, and AWS S3 cloud sync to address a common challenge in telecom network function lifecycle management

Python Pydantic Streamlit AWS S3 SQLite GitHub Actions

GitHub

🖥️ ServiceNow ITSM – Dashboards, Reports & Workflow Automation

Configured dashboards, reports and workflows to improve data-driven decision making and performed administrative tasks, validation and routine data entry for system testing on ServiceNow (Zurich)

ServiceNow ITSM Flow Designer System Testing Workflow Automation IT Operations

GitHub

🩺 Chronic Care RAG: Diabetes & Hypertension Question Answering

Built a Retrieval-Augmented Generation (RAG) medical QA system using MiniLM embeddings + FAISS with GPT-2 family models, reducing hallucinations and improving factual grounding for diabetes & hypertension queries and evaluated DistilGPT-2, GPT-2 Medium, and GPT-2 Large under baseline vs RAG settings, showing consistent gains in accuracy, contextual relevance, and response stability, with larger models benefiting most from retrieval.

Python HuggingFace FAISS MiniLM-L6-v2

GitHub

Leaf Health Semantic Segmentation Using Segformer

Built a transformer-based semantic segmentation pipeline using SegFormer-B0 for pixel-level leaf health classification, designing an end-to-end training workflow with preprocessing, augmentation, hyperparameter tuning, and CE + Dice fine-tuning to improve IoU from ~0.30 to 0.83+, while delivering quantitative (IoU, confusion matrix) and qualitative results that significantly enhanced minority-class performance.

Python HuggingFace (SegFormer-B0) Albumentations CVAT Pytorch

GitHub

⚽ FIFA Analytics Data Warehouse & Power BI Dashboard

Designed and implemented structured ETL pipelines for 200k+ records with data cleaning and validation checks, developed analytical dashboards for enterprise-style reporting, and delivered stakeholder-ready documentation and presentations translating technical findings into actionable business insights.

Python SQL Power BI Excel

GitHub Dashboard

📊 Chicago Crime Intelligence Dashboard (2021–2025)

Conducted large-scale data analysis on 1M+ records using SQL to support data-driven decision making in public administration contexts and delivered dashboards and presentations translating complex data into actionable insights to support operational efficiency and strategic vision.

Python SQL Power BI Excel

GitHub Dashboard

Credit Card Spend Revenue Analysis (2019-2020)

Analyzed 1.28M credit card transactions using Python, SQL, Excel, and Power BI to uncover revenue drivers, customer spending patterns, and seasonal trends — delivering an executive-ready 3-page interactive dashboard simulating analyst workflows at JPMorgan Chase.

Python SQL Power BI Excel

GitHub Dashboard

Surety Bond Operations Tracker

Built a 5-sheet Excel operations tracker simulating Chubb Surety's bond processing workflow — managing 20 active bonds across 4 processors, automating SLA monitoring with 169 live formulas, and reducing billing discrepancy detection time through a real-time renewal bill QA system.

Excel Dashboard Cross-Sheet Automation SLA Monitoring Renewal Bill QA Operational Workflow Training

GitHub Dashboard

Dual CNN Real-Time Dust & Stair Detection For Autonomous Vacuum

Trained two separate TensorFlow CNN models — one to classify floor patches as dusty or clean (85% accuracy), and another to classify stair, plain floor, obstacle, or unknown regions. Integrated both models into a real-time simulation using OpenCV with memory-based movement logic to avoid re-cleaning visited areas. Enabled autonomous multi-floor cleaning by detecting stairs and virtually climbing to continue cleaning across floors.

Python TensorFlow CNN OpenCV Computer Vision Deep Learning Keras

GitHub

SAP ABAP Inventory & Vendor Projects

Supported procurement and inventory business processes by developing ALV reports and SmartForms to document business processes and improve maintenance planning, assisted with data validation, vendor vetting, list configuration, and documentation to enhance reporting accuracy by 25% and ensure strong data integrity, and gained hands-on exposure to SAP ABAP modules, functional workflows, and enterprise data structures supporting effective data management.

SmartForms ALV Reports ERP Data Management SAP ABAP

MarketMindz

A university research project analyzing multi-platform digital marketing data (LinkedIn, Google, Meta) for graduate program recruitment, leveraging data analytics and AWS cloud infrastructure to evaluate campaign performance metrics like CTR and conversion rates.

AWS S3 FastAPI Uvicorn Python NLTK Regression & Classification Models

Coming Soon

Namma Nadu: A Knowledge-Grounded LLM for India's Civilizational Memory

Developing a culturally-grounded LLM for Indian historical and cultural reasoning, combining a state-wise knowledge graph, RAG architecture, and multilingual fine-tuning across India's 28 states, dynasties, oral traditions, and heritage — addressing critical gaps in existing AI models' cultural depth and regional coverage

Llama3 QLoRA Neo4j LangChain Python IndicBERT FastAPI Streamlit

Coming Soon

Things I've Built