VR.
Varshith Rajarapu
AI FLUENCY & CLAUDE 101 CERTIFIED[cite: 3]

BUILDING RESILIENT AI SYSTEMS

Computer Science engineer focused on Agentic Integrations and Full-Stack Architectures[cite: 1, 2, 3]. From real-time fraud detection pipelines to LLM-powered analytics, I build tools that translate technical complexity into organizational impact[cite: 1, 3, 5].

EXPLORE SYSTEMS GET IN TOUCH
01 / SUMMARY

THE UTILITY PLAYER

I am a versatile software engineer with hands-on experience developing end-to-end applications and automation tools[cite: 1, 3]. My approach combines systems-level thinking with a user-centered focus, ensuring that data-driven insights are actionable for non-technical stakeholders[cite: 1, 3, 5].

Whether it’s optimizing PostgreSQL schemas for 44% faster load times or orchestrating LLM pipelines that cut time-to-insight by 70%, I thrive in environments where resource constraints meet high-stakes requirements[cite: 1, 3, 4, 5].

91%
ML PRECISION[cite: 5]
82%
MODERATION EFFICIENCY[cite: 2, 5]
70%
FASTER ANALYTICS[cite: 1]
[EDUCATION]
B.S. Computer Science[cite: 5]
UW-Eau Claire (GPA: 3.5)[cite: 4, 5]
[CERTIFICATIONS]
Claude 101 (Anthropic)[cite: 3]
AI Fluency (Anthropic)[cite: 3]
[CORE TOOLS]
Claude Code, Cursor, GitHub Actions, Docker, AWS (S3/RDS)[cite: 1, 2, 3]
02 / WORK

FEATURED SYSTEMS

FIN.AILLM PIPELINE[cite: 1]

FinTrack Analytics[cite: 1, 3]

Orchestrated an LLM pipeline with OpenAI API to generate natural-language spending summaries from multi-account transaction data[cite: 1, 3].

OpenAI API[cite: 1, 3]Next.js[cite: 3]PostgreSQL[cite: 1, 3]Pytest[cite: 1]
CHAT.MODWEBSOCKETS[cite: 2, 3]

AI Adaptive Chat[cite: 2, 3]

WebSocket-based chat system handling 300+ users with <50ms delivery[cite: 2, 3]. Integrated inline LLM moderation to auto-flag toxicity[cite: 2, 3, 5].

WebSockets[cite: 2, 3]Node.js[cite: 2, 3]TypeScript[cite: 2, 3]Redis[cite: 2]
DOC.FLOWCIVIC TECH[cite: 3]

DocFlow Civic Suite[cite: 3]

Secure document verification interface for cross-agency discrepancies[cite: 3]. Reduced review cycles by 38% via custom annotation tools[cite: 1, 3].

GraphQL[cite: 2, 3]AWS S3[cite: 2, 3]Docker[cite: 2, 3]React[cite: 3]
MED.SYNCWORKFLOW AUTOMATION[cite: 4, 5]

PatientFlow System[cite: 4, 5]

Clinical intake engine managing hundreds of organizational rules[cite: 4, 5]. Automated reminders reduced absenteeism by 31%[cite: 4, 5].

C# .NET[cite: 4, 5]Spring Boot[cite: 4, 6]User-Centered Design[cite: 4, 5]
03 / RESEARCH

Ad Fraud Detection Pipeline.[cite: 3, 5]

Engineered a machine learning pipeline at UW-Eau Claire to classify invalid ad traffic across 2M+ data points[cite: 4, 5]. Using Random Forest and XGBoost, I achieved 91% precision[cite: 4, 5]. I translated these complex ML outputs into a real-time FastAPI-powered dashboard for non-technical faculty stakeholders[cite: 3, 5].

Python[cite: 2, 3]Pandas[cite: 2, 3]Scikit-learn[cite: 2, 3]FastAPI[cite: 2, 3]Matplotlib[cite: 2, 3]
91%
MODEL PRECISION[cite: 4, 5]
```