Open to Software Engineering and AI/ML roles

Hi, I'm

Aruthra Sathish Kumar

Software Engineer

Backend & Distributed Systems · Applied AI / ML

I build scalable backend systems and intelligent applications.

Get in Touch View Projects

GitHubLinkedIn

About

Software Engineer and AI/ML Researcher

Recent M.S. graduate from George Mason University with a 3.97/4.0 GPA, recognized with the Academic Excellence Award (2026). I enjoy building scalable backend and distributed systems, while exploring how AI and machine learning can make software more intelligent.

Currently, I'm working on AI/ML research at American University, where I work with large language models and applied machine learning.

I love solving problems where software, scale, and intelligence come together. Outside of coding, you'll probably find me playing chess ♟️, exploring new technology, or getting completely absorbed in learning something new.

Backend & Distributed Systems

Build scalable, reliable systems.

Applied AI & Machine Learning

Build intelligent systems with AI.

Cloud & Systems

Design infrastructure that scales.

Education

Academic Background

The foundation behind my work in software and AI/ML engineering.

Master of Science - Information Systems

George Mason University

Aug 2024 - May 2026

3.97 / 4.00GPAAcademic Excellence AwardRecognition

Graduate program focused on backend systems, distributed computing, data engineering, and AI/ML applications.

Projects in distributed systems, RAG platforms, and cloud-native services
Emphasis on software architecture, data modeling, and cloud infrastructure

Bachelor of Technology - Information Technology

Anna University

Aug 2020 - May 2024

Undergraduate foundation in algorithms, data structures, networking, databases, and full-stack development.

Experience

Where I've Built

Research, Teaching, and Production Engineering Expertise

AI / ML Research Intern

Current

American University

Jul 2026 - PresentMachine Learning / NLP Research

datasets harmonized

parameter LLM adapted

representation paths fused

CAM-Soft: scalable multimodal fusion for soft hate speech detection via camouflaged-implicature modeling and counterfactual invariance.

ARCHITECTED dual-pathway hate speech detection in PyTorch → harmonized 147K posts across seven benchmarks.
FINE-TUNED NV-Embed-v2 with PEFT LoRA → achieved 0.835 macro-F1 across five benchmarks.
BUILT cached Llama-3.1-8B reasoning pipeline → extracted pragmatic features from 140K+ posts.
ENGINEERED implicature gap & camouflage scoring → modeled differences between literal and implied meaning.
FUSED 6 semantic & pragmatic signals with multi-head cross-attention → improved macro-F1 from 0.7504 to 0.7854.
DEVELOPED a dynamic coded-term inventory → tracked emerging coded language across communities and time.

Graduate Teaching Assistant

George Mason University

IT 207 - Applied IT Programming (Server-Side Development)

Aug 2025 - May 2026Backend Engineering Instruction

students supported

higher lab completion

fewer runtime failures

Taught server-side development, REST API design, and SQL to undergraduate students.

MENTORED 100+ students on Node.js, SQL, and distributed systems → improved lab completion & pass rates.
TRAINED students to build SQL-based backend services → 30% fewer runtime failures.
LED weekly coding & debugging labs on RESTful service design → enabled students to build backend services.

Software Engineer Intern

Verzeo Edutech

Dec 2022 - Jan 2023Full-Stack Engineering

fewer auth issues

fewer DB round trips

fewer client calls

Built and optimized full-stack features on an e-learning platform serving 1K+ students.

BUILT full-stack features (React + Django) → supported 1K+ students accessing courses & assessments smoothly.
DESIGNED focused REST API layer for course/progress data → 40% fewer client calls & 25% faster page loads.
RESOLVED login failures with OAuth2 + OpenAthens SSO → 45% fewer auth issues during peak hours.
OPTIMIZED relational schemas for courses, users & attempts → reduced heavy query latency by 30–45%.
STANDARDIZED cursor pagination & server-side joins in REST layer → 60% fewer DB round-trips during peak traffic.
COLLABORATED with engineers via code reviews & Git workflows → improved API consistency & codebase quality.

Skills

Technical Strengths

HTML/CSS90%

JavaScript90%

ReactJS95%

TypeScript85%

React Native80%

TailwindCSS80%

Next.js85%

Python90%

FastAPI85%

Java85%

REST & WebSockets84%

PostgreSQL82%

Redis82%

Node.js80%

Kafka78%

MySQL76%

Socket.io76%

C / C++75%

Fastify74%

Apache Flink72%

MongoDB68%

Django66%

Prompt Engineering85%

Natural Language Processing (NLP)85%

Machine Learning (General)85%

RAG pipelines84%

LLM integration82%

Pandas & NumPy82%

PyTorch80%

Embeddings & FAISS80%

AI Agents (Multi-Agent Systems)80%

scikit-learn78%

Hugging Face76%

MCP & pgvector76%

Speech & translation APIs72%

LoRA / PEFT70%

MLflow66%

Git & GitHub86%

Docker80%

GitHub Actions & CI/CD78%

AWS74%

Kubernetes72%

Azure70%

Prometheus & Loki70%

Terraform68%

Projects

Featured Projects

AI Infrastructure

PythonMCPPostgreSQL + pgvectorKubernetes+3

WatchTower - MCP Server for Incident Response

Claude Desktop drives 12 MCP tools to triage and safely remediate Kubernetes incidents.

Claude Desktop → MCP Server → Collectors → Event Store → Suspect Ranking → Approval Broker → K8s Remediation

faster incident triage

MCP tools exposed

live data sources

Code

Real-Time Search Ranking System project cover

Distributed Systems

Node.jsKafkaApache FlinkRedis+2

Real-Time Search Ranking System

Reorders search results in real time from live clickstream data.

Clickstream → Collection API → Kafka → Flink → Redis → Search API

1K+

events/min processed

<50ms

search latency

30%

relevance lift (CTR A/B)

Code

USDA Rural Development AI Assistant project cover

RAG System

ReactTailwindFastAPIPython+4

USDA Rural Development AI Assistant

Natural-language search across 176 USDA programs with citation-backed answers.

User Query → BGE Embeddings → FAISS → Retrieved Programs → Mistral 7B → Grounded Answer

176

federal programs indexed

90s to 8s

response time

100%

answers grounded in retrieval

Code

Real-Time Systems

Next.js 14FastifyPostgreSQLRedis+4

SpeakUp - Anonymous Voice Q&A Platform

Multilingual anonymous Q&A for live events, with voice questions and live voting.

Browser → Fastify API → Kafka → Whisper / DeepL → Postgres + Redis → Socket.io

500+

concurrent users per room

<100ms

vote sync latency

99+

languages transcribed

Code

Academic Performance Predictor project cover

Deep Learning

PyTorchScikit-learnPandasMLflow+1

Academic Performance Intelligence System

Predicts student risk and grades from LMS activity with a multi-task CNN-GRU.

LMS Data → Feature Pipeline → CNN → GRU + Attention → Risk + Grade

88.9%

risk prediction accuracy

59%

lower error vs baseline

Code

Backend Systems

ReactFastAPIRedisMySQL+2

Pathly - URL Shortener

URL shortener with cache-aside redirects and asynchronous click analytics.

Client → FastAPI → Redis → MySQL → Analytics

<1ms

cached redirects

100%

clicks tracked async

Code

Full-Stack

ReactFastAPIPostgreSQLJWT+3

CareerLens - Job Search Intelligence Platform

Turns application tracking into funnel analytics, fed by a Chrome extension.

Chrome Extension → FastAPI → PostgreSQL → Analytics API → Dashboard

60%

less manual entry

<1s

SQL funnel aggregations

Code

View all repositories

Awards

Awards & Recognition

Recognition

Academic Excellence Award

George Mason University · 2026

Recognized for outstanding academic performance in the M.S. Information Systems program at George Mason University.

Cumulative GPA: 3.97 / 4.00
Program: M.S. Information Systems
Year awarded: 2026

Credentials

Certifications

AWS Academy - Data Engineering

TRAINEDAmazon Web Services · December 2024

AWS CloudData PipelinesETL

Verify credential

Microsoft Certified - Azure Fundamentals (AZ-900)

CERTIFIEDMicrosoft · February 2025

AzureCloud FundamentalsSecurity

Verify credential

Contact

Get in Touch

Open to backend, distributed systems, infrastructure and applied AI/ML roles. Say hello.