Rindhuja Johnson | Computational Physicist and Data Scientist

About Me

Computational Physicist and Data Scientist

MPS in Data Science University of Maryland - Baltimore County, USA | 4.0 GPA

M.Sc. Physics Pondicherry University, India | 3.61 GPA

I'm a Physicist with expertise in Python programming and numerical modeling. My experience extends to AI Development, specializing in multi-agent AI systems, LLM optimization, and enterprise-grade infrastructure. I build intelligent systems that solve complex problems at scale.

My journey from theoretical physics to cutting-edge AI development has equipped me with a unique perspective on problem-solving. At UMBC, I discovered the transformative power of data during my undergraduate internship analyzing El-Niño Southern Oscillations at CUSAT.

I am passionate about using my unique skill sets in Physics and data to bridge the gap is slowing down numerical modeling and computational experiments from yielding result.

Core Competencies

Physics Data Analysis Python Programming Prompt Engineering Cloud Architecture (GCP/AWS) Token Budget Management RAG Pipelines RLHF & SFT

Kochi, India (Willing to Relocate)

Phi Kappa Phi Honor Society Member

Data Writer @ Medium

AI Expertise

Specialized in Multi-Agent Orchestration & LLM Optimization

AI Systems Architecture & Development

Built enterprise-grade multi-agent systems, LLM optimization, and cloud infrastructure

Multi-Agent Systems Design

Expertise in designing and implementing collaborative AI agent systems:

Agent Orchestration: Coordinate multiple agents for complex workflow execution
Task Routing: Intelligent request analysis and distribution to specialized agents
Data Processing: Iterative data collection with real-time progress tracking
Insight Generation: Automated synthesis of findings into actionable recommendations
Enterprise Integration: Security, compliance, and third-party system connectivity

Python FastAPI WebSockets Claude API

LLM Cost Optimization

Proven track record of optimizing AI operations for cost and efficiency:

Workflow Analysis: Identify prompt inefficiencies and excessive context retrieval
Budget Management: Implement token/cost limits with configurable enforcement
Model Selection: Optimize routing between different model tiers (Opus/Sonnet/Haiku)
Caching Strategies: Design intelligent caching for frequently-used prompts
Performance Monitoring: Track and prevent redundant API calls

Achieved 7-10K token savings per session in production

AI Workflow Orchestration

Building scalable systems for AI task management:

Parallel Processing: Execute multiple AI workflows concurrently with resource management
Smart Scheduling: Time-based task execution for optimal resource utilization
Resource Controls: Implement budget limits and usage quotas per workflow
Live Monitoring: Real-time dashboards showing progress, costs, and performance
Event Architecture: WebSocket-based communication for transparent operations

asyncio Event Streaming Task Scheduling

Cloud & DevOps

Deploying and managing enterprise-grade cloud infrastructure:

Serverless Computing: Cloud Run, Lambda with auto-scaling and cost optimization
Database Management: PostgreSQL, Redis for high-availability data storage
Infrastructure as Code: Terraform for version-controlled, reproducible deployments
Containerization: Docker for consistent environments across dev and production
Monitoring: OpenTelemetry, distributed tracing, centralized logging

GCP AWS PostgreSQL Docker Terraform

LLM Integration

Building robust multi-provider LLM applications:

API Integration: Claude, GPT-4, Gemini with unified interface
Reliability: Timeout protection, circuit breakers, and graceful degradation
Error Handling: Exponential backoff and automatic retry for transient failures
Usage Analytics: Track token consumption, costs, and performance metrics
Model Management: Dynamic routing between providers based on requirements

Anthropic API OpenAI API Gemini API LangChain

Backend Development

Building high-performance Python backend systems:

API Development: FastAPI, Flask with async/await for high concurrency
Database Layer: SQLAlchemy ORM, asyncpg for PostgreSQL, Redis for caching
Security: JWT authentication, OAuth2, password hashing (argon2)
Data Processing: Pandas, NumPy for ETL and analytics pipelines
Observability: Prometheus metrics, structured logging, distributed tracing

FastAPI SQLAlchemy Pandas JWT

Additional Expertise

Data Science & Analytics

Statistical modeling & predictive analytics
Power BI, Tableau visualization
A/B testing & customer segmentation
180% engagement improvements via analytics

ML & Deep Learning

PyTorch, TensorFlow, Keras
Feature engineering & pattern analysis
95% model accuracy achievements
Time-series forecasting (SARIMAX)

RAG & Vector Search

Retrieval-Augmented Generation pipelines
LangChain, Ollama, Chroma DB
Real-time query resolution systems
Bing Search API integrations

LLM & Prompt Engineering

Prompt optimization techniques (5+ strategies)
RLHF initiatives & SFT implementations
Enhanced AI accuracy by 25-30%
LaTeX-formatted LLM responses

Big Data & Cloud

Apache Spark, Hadoop, Databricks
GCP, AWS, Azure infrastructure
PostgreSQL, Redis, Cloud SQL
7GB+ dataset processing

Professional Experience

Quality Control Auditor (Physics)

Scale AI

Nov 2025 - Present

Evaluated high-complexity physics problems spanning classical mechanics, electromagnetism, quantum mechanics, statistical mechanics, and mathematical physics

Conducted domain-level audits of AI-generated scientific responses, verifying conceptual correctness, mathematical rigor, and methodological validity, and provided constructive feedback to the contributing experts

Demonstrated accountability to the audited tasks by ensuring all the issues in the prompt-response pair is addressed and the contributing experts are guided to perfection

Collaborated cross-functionally with ML engineers and team leads to refine domain-specific evaluation rubrics

Proposed changes in quality assurance frameworks grounded in scientific validation principles rather than heuristics

Physics Research Quality Prompt Engineering Math

Lead AI Developer

Netra Systems Inc.

Aug 2025 - Nov 2025

Developed AI optimization backend achieving 7-10K token savings per session through automated workflow analysis and metadata diagnostics

Architected multi-agent orchestration system with real-time token monitoring, cost tracking, and configurable budget enforcement

Built enterprise pricing engine providing transparent model-routing and cost analysis for Claude Opus/Sonnet/Haiku

Automated cloud infrastructure deployment using Terraform, Docker, GCP Cloud Run - reducing deployment time by days

Implemented distributed backend services with FastAPI, WebSockets, JWT auth, OpenTelemetry for high-performance operations

Python FastAPI Claude API PostgreSQL Redis GCP Docker Terraform

Data Scientist

Outlier (Scale AI)

Aug 2024 - Present

Enhanced AI model accuracy by 25% through feature engineering and pattern analysis

Achieved 95% model accuracy via advanced QA frameworks and validation protocols

Improved AI-generated content accuracy by 30% using Supervised Fine-Tuning (SFT)

Led RLHF initiatives, mitigating loss categories and refining model performance

Prompt Engineering RLHF SFT QA Frameworks LLM Optimization

Junior Data Scientist

Leveragai Inc.

Aug 2024 - Jul 2025

Led development of AI Tutor for Generative AI course using Azure OpenAI

Designed RAG pipeline for real-time query resolution with LaTeX-formatted responses

Optimized prompt engineering strategies improving AI accuracy by 30%

Azure OpenAI RAG LangChain Bing Search API

Data Analyst

Redwood Algorithms

Jan 2022 - Dec 2022

Increased customer engagement by 180% through predictive analytics

Boosted ad campaign ROI by 25% through CTR analysis

Python SQL BigQuery Power BI

Featured Projects

Computational Physics, Data Analysis, Visualization

Featured

Geomagnetic Field Modeling and Atmospheric CR Induced Ionization Analysis through Ages

Computes the global geomagnetic rigidity cutoffs using paleomagnetic field models (LSMOD.2, OTSO, CRAC-CRII-v3) and obtains atmospheric ionization and effective dose rates to compare present-day vs excursion scenarios, over latitude and altitude

Geomagnetism Physics Python Numerical modeling LSMOD.2 OTSO CRAC-CRII-v3

View on GitHub

Featured

AI Workflow Optimization System

Open-source tool for optimizing AI developer workflows with Claude Code, Gemini, and Codex. Implements parallel orchestration, intelligent scheduling, real-time cost tracking, and automated optimization achieving 7-10K token savings per session.

Python WebSocket Multi-Agent Systems LLM APIs Cost Analytics

View on GitHub