[ ABOUT ]

Building intelligent systems

Crafting AI that makes it past the notebook.

Final-year AI/ML Engineer specializing in computer vision, NLP, RAG systems, and AI agent development. Proficient in Claude AI for intelligent code generation and agentic workflows.

Built production-grade solutions — real-time anomaly detection and RAG pipelines. Microsoft Hackathon Finalist with a strong foundation in deep learning, model optimization, and MLOps.

6
Months
6+
Projects
8.5
CGPA
MSFT
Hackathon Finalist

Computer Vision

YOLOv8, YOLO-World, Grounding DINO, OpenCV — real-time detection systems.

NLP & RAG

Transformers, FAISS, LangChain, reranking — semantic search that returns the right thing.

AI Agents

Multi-agent systems, function calling, n8n automation — agents that take action.

ML Engineering

Deep learning, model optimization, MLOps, Docker + Kubernetes.

[ SKILLS & TECH ]

The stack I work with

Modern tools for production-grade AI systems.

Programming

PythonSQLC++JavaScriptTypeScript

AI / LLM

Claude APILangChainLangGraphAgent DevelopmentFunction CallingMulti-AgentAutomation

ML & AI

SupervisedUnsupervisedDeep LearningFeature EngineeringOptimization

Computer Vision

YOLOv8YOLO-WorldGrounding DINOOpenCV

NLP & RAG

TransformersHugging FaceRAGFAISSEmbeddings

Frameworks

TensorFlowPyTorchScikit-learnXGBoostPandasNumPy

Backend

FlaskREST APIsMongoDBMySQLn8n

DevOps

DockerKubernetesCI/CDGit

Analytics

Power BIExcelMatplotlibSeaborn

[ EXPERIENCE ]

Where I've shipped

Real-world AI work at scale.

AI Developer Intern

Internship

Codeacious Technologies

Dec 2025 – May 2026Remote
  • Built real-time Event Anomaly Detection using YOLOv8, YOLO-World, Grounding DINO
  • Developed production RAG pipelines with retriever + reranker architecture
  • Created backend using NestJS, React, MongoDB
  • Deployed using Docker & Kubernetes
  • Automated workflows using n8n
1000+
Events/min
95%
mAP Accuracy
40%
RAG Improvement
PythonYOLOv8YOLO-WorldGrounding DINONestJSReactMongoDBDockerKubernetesn8nFAISSRAG

[ FEATURED WORK ]

Projects that ship

Production-grade AI/ML systems — computer vision, NLP, RAG, and agent orchestration.

AI Surveillance System

Internship

Real-time video anomaly detection with open-vocabulary models. Scalable backend, sub-second alerts, dockerized for prod.

YOLOv8YOLO-WorldGrounding DINODocker

DocMind-RAG

AI / NLP

Document Q&A with hybrid semantic search and cross-encoder reranking. Cuts hallucinations, returns citation-grounded answers from PDFs.

FAISSRerankersTransformersEmbeddingsHuggingFace
CodeDemo

Portfolio Optimization ANN

Deep-learning asset allocator. 23% returns on backtest, 0.85 Sharpe, live demo with real-time analysis.

TensorFlowKerasStreamlit
CodeDemo

Movie Genre NLP

Plot-to-genre classifier across 20+ genres. 94% accuracy with classical ML pipeline and bag-of-words features.

Scikit-learnNLPStreamlit
CodeDemo

Course Recommender

Personalized learning-path engine. Indexes 10K+ Udemy courses, matches by intent, surfaces relevant tracks fast.

NLPScikit-learnStreamlit
CodeDemo

HR Sentinel Automation AI

Enterprise

Enterprise HR automation. Processes biometric attendance logs, detects late arrivals, and triggers AI-powered escalations via Gmail and Google Calendar.

n8nGPT-4oPostgreSQLReactPython
CodeDemo
View all on GitHub

[ ACHIEVEMENTS ]

Highlights & recognition

Wins, milestones, and work I'm proud of.

Microsoft Hackathon Finalist

2025
Selected among thousands for building an AI-powered productivity solution.

AI Developer Intern — Codeacious

2025–2026
Shipped real-time anomaly detection and production RAG pipelines.

8.5 CGPA — Final-year CS undergrad

2022–2026
Strong theoretical grounding in algorithms, ML, deep learning, and systems.

Open-source projects

Ongoing
Portfolio Optimization ANN, Movie Genre NLP, Course Recommender — all live.

[ CONTACT ]

Let's build together

Hiring, collaborating, or just want to talk shop about AI? Inbox is open.

Email

harsh078976@gmail.com

Phone

+91 8384030533

Location

India

Ready to start something?

Drop me a line — I usually reply within a day.

Send a messageGitHubLinkedIn