MCA AI/ML student focused on shipping applied AI systems beyond notebooks: retrieval pipelines, ML APIs, evaluation workflows, and deployment-ready backends.
I build projects with clear problem framing, clean architecture, tests, documentation, and production-oriented engineering habits.
- RAG systems with citations, retrieval tracing, evaluation, and deployment readiness
- NLP pipelines for classification, clustering, routing, and support automation
- GenAI backends using FastAPI, TypeScript, streaming, memory, and external APIs
- ML system design with CI, Docker, model cards, dataset cards, and deployment docs
| Project | Impact | Tech | Link |
|---|---|---|---|
| TraceRAG | Full-stack RAG system with document ingestion, hybrid retrieval, citation grounding, ACL, query tracing, evals, CI, Docker, and Next.js console. | FastAPI, Next.js, Postgres, pgvector, OpenAI, Python, TypeScript | Repo |
| DocuMind AI Copilot | Customer-support RAG copilot that turns uploaded policy PDFs into citation-backed answers with memory, reranking, and streaming responses. | FastAPI, FAISS, BM25, OpenAI, Python | Repo |
| Customer Inquiry Classifier | Confidence-aware NLP routing system that classifies support messages and escalates uncertain cases for human review. | scikit-learn, FastAPI, Streamlit, Python | Repo · Demo |
| Document Clustering and Topic Modeling | Unsupervised NLP pipeline for grouping documents, extracting themes, evaluating cluster quality, and exploring results interactively. | scikit-learn, NLTK, Streamlit, Python | Repo · Demo |
| AI Trip Planner Frontend | A sophisticated AI-frontend integration that streamlines complex trip planning into instantaneous, data-driven, and personalized travel itineraries. | React, Vite, Tailwind CSS, Lucide React | Repo · Demo |
AI/ML: RAG, NLP, text classification, clustering, topic modeling, evaluation, model cards
Backend: FastAPI, Node.js, Express, REST APIs, SSE, authentication, rate limiting
Data and Retrieval: Postgres, pgvector, FAISS, BM25, SQLAlchemy, Alembic
Frontend: Next.js, React, TypeScript, Streamlit
MLOps: Docker, GitHub Actions, deployment docs, env templates, tests, CI gates
Languages: Python, TypeScript, JavaScript, SQL
- Reliable RAG systems with measurable retrieval quality and traceable answers
- NLP tools that solve operational problems such as support routing and document discovery
- GenAI backends that are testable, deployable, and understandable to senior engineers
- A portfolio that demonstrates engineering judgment, not just model usage
- GitHub: github.com/Yashsh101
- LinkedIn: linkedin.com/in/yash-sharma-262923183
- Portfolio: yashsharma01.vercel.app
- Resume: Yash-Sharma.pdf
- Email: [email protected]