E-commerce search is broken – filters are clunky, results are irrelevant.
‍Built a 10X better search engine – LLM-powered and agentic. Perplexity for shopping  
Semantic search - Uses Vector embedding along with LLM to provide Best Result  
‍Understands natural language – just type “T-shirt for a music concert,” and it finds exactly what you need.
‍Scrapes 12 top fashion sites – brings back the best results instantly.
‍No more endless scrolling – search smarter, shop faster.


TECHNOLOGY USED: OpenAI api, FastAPI, Vector Embedding,
Pinecone , selenium , Flask

Agentic Workflow for CIS Benchmark Audit & Remediation – The AI agent splits PDFs into bite-sized sections to extract only the essential data.
Script Generation – Relevant sections are sent to an LLM that crafts tailored audit and remediation scripts aligned with CIS benchmarks.
Rigorous Quality Assurance – A secondary LLM meticulously reviews the scripts to identify any issues.
Instant Refinement – Detected issues are immediately sent back to the LLM for precise fixes.
Execution & Analysis – The finalized scripts are executed and their outputs are closely analyzed.
Continuous Optimization – Any anomalies trigger further LLM iterations until the scripts are flawless.

Problem Overview – Many organizations face challenges when non-technical staff must run command-line scripts without a GUI, and acquiring CIS hardened virtual images can be prohibitively expensive
CIS Benchmark Audit Automation – Our automation scripts conduct comprehensive security audits aligned with CIS Benchmarks, ensuring robust compliance and enhanced security.
Seamless GUI Execution – A PySide6-based GUI auto-detects the OS and runs the appropriate scripts.
Efficient Reporting & Search – Audit results are exported as PDFs, with a built-in search engine and structured database for quick script access.

Know My College – A Hackathon-winning webapp that simplifies finding the right college using just 10 straightforward questions. The app gathers your key preferences—such as job opportunities, campus life (rated on a scale of 1 to 100), and your preferred states—and returns a personalized, sorted list of colleges.
Personalized Recommendations – Leveraging a Pandas DataFrame , the app normalizes various parameters with linear regression and applies a weighted average algorithm to generate college recommendations tailored to your unique criteria.
College Insights – A RAG pipeline combined with a vector database powers a chatbot that delivers comprehensive, pre-acquired college information, including details on placements, campus facilities, and other key metrics to help you make an informed decision.