Building Intelligent Systems with AI & Machine Learning
AI Engineer focused on Machine Learning, Deep Learning, LLMs, Computer Vision, and Generative AI. I build things that work in production, not just in notebooks.
From curiosity to intelligent systems
The story behind the work - how I got into AI, what drives me, and where I'm headed.
I'm a B.Tech Computer Science student at Rai University, Ahmedabad, with a focus on AI and machine learning. I spend most of my time building projects that actually run in production, not just notebooks.
My interest in ML started when I wanted to understand how machines pick up on patterns that humans miss. That pulled me into Python, linear algebra, and frameworks like Scikit-Learn, TensorFlow, and PyTorch. I've since built classifiers, neural networks, and LLM-powered apps, and I care deeply about whether a model works in the real world, not just on a test set.
Right now I'm focused on LLMs, Retrieval-Augmented Generation, and applied deep learning. I'm looking for an AI/ML internship where I can work on real production systems alongside people who care about the craft.
Why ML
I want to understand how machines find patterns that humans can't. The engineering required to make those patterns useful is what keeps me going.
Research interests
LLMs, Retrieval-Augmented Generation, applied deep learning, and the gap between a good offline metric and actual user impact.
Career goal
An AI/ML internship where I can contribute to real production systems and learn from engineers who've shipped at scale.
My journey into AI
Started B.Tech in Computer Science
Rai University, Ahmedabad
Began CS degree focusing on Data Structures, Algorithms, Statistics, and Databases. Built practical ML and AI foundations from scratch.
Artificial Intelligence Intern
Codec Technologies India (Remote)
Developed ML models and NLP pipelines with Scikit-Learn, Pandas, and NumPy. Trained classification models (XGBoost, SVM, Random Forest) with 90%+ accuracy and containerized inference REST endpoints with FastAPI and Docker.
Agentic AI Intern
Codage Habitation, Ahmedabad (On-site)
Architecting enterprise Agentic AI systems with Python and LangGraph. Building execution graphs with dynamic runtime nodes, MCP tool bindings, and high-precision RAG pipelines with 0.65 relevance gates.
Autonomous Agentic AI Platforms
AegisCode, LeadFlow AI & Agentflow-AI
Architected autonomous multi-agent execution graphs, GraphRAG code intelligence, and enterprise sales/operations platforms orchestrating Groq LLaMA 3.3, Twenty CRM, Zapier, and MCP tools.
An ML toolkit, end to end
From the math and the models to the frameworks and the deployment - the stack I use to take ideas from notebook to production.
Capability profile
Relative strength across core AI/ML competencies
AI & Machine Learning
Frameworks & Libraries
Programming
Cloud & DevOps
Projects I've built
Real projects - from autonomous multi-agent systems and ML classifiers to full-stack production platforms. Each ships with a live demo, source code, and a full case study.
Autonomous multi-agent execution graphs, LangGraph state machines, MCP tool execution, and code intelligence platforms.
Enterprise-grade autonomous AI software engineering platform with multi-agent orchestration, sandboxed execution, and security guardrails.
Autonomous B2B sales operations platform orchestrating lead capture, qualification, Twenty CRM, and Zapier.
Enterprise-grade autonomous Git debugging and code intelligence platform powered by GraphRAG and a 28-state engine.
Production-grade autonomous enterprise operations copilot implementing 'Ask, Retrieve, Act' with LangGraph and MCP.
A comprehensive AI workspace featuring 6+ specialized modules powered by Groq.
AI-powered technical assistant for code generation and debugging.
Evaluation metrics, end to end
Every model is only as good as its evaluation. Here are the headline metrics for my trained models - accuracy, precision, recall, F1, and the ROC profile.
ROC profile · AUC 0.940
Brain Tumor CNN
Deep Learning · 4 classes · TensorFlow
Area Under ROC
0.940
Excellent discrimination
Try my models - live
Run real inference against three of my trained models. Upload a file, paste text, and watch the pipeline execute step by step.
Quick examples
Paste text and click Run Inference
How my stack connects
An interactive map of the tools I work with and how they relate - from Python and PyTorch to LLMs, vector stores, and cloud. Hover a node to trace its connections.
← Scroll to explore the full graph →
Always learning, always verifying
Workshops and trainings from my LinkedIn profile - each one verifiable with a direct link.
Industry validation and opportunities
A collection of internship offers I've received, validating my skills in AI, Machine Learning, and Full-Stack Development.
I build in public
A snapshot of my GitHub activity - contributions, top languages, and streaks. Code is the resume that can't be faked.
Contribution graph
@HarshPariyaWhat people say
Recommendations from mentors, teammates, and collaborators.
"Harsh builds things end to end. He doesn't just run experiments in notebooks - he gets the model into a UI and deploys it. That kind of follow-through is rare at this stage."
Let's build something intelligent
Open to AI/ML internships and research roles. If you have a project, a role, or just want to talk ML - reach out.