Arav Bhardwaj

Chief Technology Officer & Full-stack Engineer

Building intelligent systems at the intersection of AI and software engineering. Currently studying AI and Statistics & Machine Learning at Carnegie Mellon University, and leading technical development at Basics.

About

I'm a developer and entrepreneur passionate about building intelligent systems that solve real-world problems. Currently pursuing a BS in AI and Statistics & Machine Learning at Carnegie Mellon University (Expected May 2028, GPA: 3.5), where I'm focusing on creating scalable AI-powered applications that make technology more accessible and intuitive.

As Chief Technology Officer at Basics, I'm leading the development of an AI-powered learning platform that validates MLOps work completion and automates training content creation. Previously, I've built enterprise AI solutions at Karini AI, developed intelligent recruiting tools at Sparo, and conducted medical AI research at Dartmouth-Hitchcock Medical Center.

I specialize in full-stack development with Next.js, React, and Python, alongside deep experience in AI/ML frameworks like LangChain, LangGraph, PyTorch, and LLM integrations. My technical expertise spans from building conversational AI systems and computer vision applications to scalable backend infrastructure with Go, Node.js, and cloud platforms like AWS and Azure.

Experience

Chief Technology Officer

·Basics
2024 — Present
  • Developed real-time monitoring infrastructure that tracks MLOps execution across development environments, leading to a $5K pilot customer acquisition in under 2 months
  • Architected automated content generation system using LangGraph that transforms unstructured company data into validated workflows, cutting content creation time from 40+ hours to under 2 hours
  • Drove early-stage growth through rapid product iteration and customer discovery, reaching YC final-round interviews and securing 1st place at Pittsburgh pitch competition
LangGraphNext.jsTypeScriptPythonMCPMLOps

Fullstack Software Engineer

·Karini AI
May 2025 — July 2025
  • Created MCP Registry™ platform managing governance for 30+ production servers across 10+ enterprise clients, streamlining security deployments from days to minutes
  • Architected Microsoft Teams AI integration for enterprise client processing 1,000+ documents with sub-2 second response times and 95%+ accurate citations
  • Enhanced multi-agent system reliability through A2A protocol integration and Ragas evaluation framework implementation in LangGraph architecture
Next.jsReactMongoDBAzureBot Framework SDKLangGraphRagas

Backend Software Intern

·Sparo
June 2024 — July 2024
  • Designed and implemented automated LinkedIn data extraction system processing 1,000+ profiles at 120 profiles/hour with 98% accuracy, eliminating 85% of manual research overhead
  • Developed intelligent matching algorithms leveraging connection mapping to identify networking opportunities, boosting candidate placement success rate by 30%
Node.jsPuppeteerJavaScriptData Processing

Machine Learning Intern

·Dartmouth-Hitchcock Medical Center
June 2023 — July 2023
  • Designed Wasserstein GAN architecture with gradient penalty for generating synthetic colorectal tissue images from omics data, using spectral normalization to improve training stability
  • Optimized HPC workflows on Slurm clusters for whole slide image processing, achieving 40% performance improvement through distributed parallel processing of terabyte-scale datasets
PyTorchPythonGANsHPCSlurmComputer Vision

Projects

Featured

Sentinel

View Code
  • Go-based AI governance gateway preventing cost overruns across LLM providers and MCP servers
  • Reduced policy deployment from hours to under 5 minutes via Python SDK
  • Policy engine with model-specific, provider-wide, and global enforcement
GoPythonSQLiteLangChainMCP

Verity

View Code
  • Content authenticity platform with AI detection, IPFS storage, and Polygon smart contracts
  • Full-stack app with FastAPI backend and Next.js frontend for verification
  • Immutable verification system for digital content tampering and deepfakes
PythonFastAPINext.jsMongoDBIPFSPolygon

Ecosort

View Code
  • Autonomous waste sorting with 82% accuracy using YOLOv5 on 25K+ images
  • Multi-threaded architecture optimized for 30 FPS on Raspberry Pi 5
  • Precision servo-controlled mechanical separation system
PythonYOLOv5OpenCVRaspberry PiComputer Vision

Liemap

View Code
  • NetLogo simulation visualizing the spread of lies in different cultural contexts
  • Analyzes deception patterns based on severity, benefit, and social proximity
NetLogoAgent-Based Modeling

PyScan Pro

View Code
  • Learning tool with integrated code editor and step-by-step tracing
  • Comprehensive Python programming education platform
PythonOpenCVPILTesseract

Contact

Get in touch

Have a project in mind or just want to chat? Send me a message below.