Things I’ve builtat work, in hackathons and at university.

Vimo · work project

Regression failure-management system

Turns a night of failed regression tests into assigned, tracked Jira defects.

Impact: Failure distribution went from 3-4 hours of manual work to 5-10 minutes, for 50-60 nightly failures across 9 owners.

• Pulls run results from the Jenkins and Allure APIs, on a schedule or on demand.
• Finds an owner for each failure in four steps: scenario-author mappings, Git history, keywords in the failed step, then feature-directory rules. Normalizes Git identities, and uses round-robin when nothing matches.
• Stores results per state environment in PostgreSQL and shows them on an internal dashboard: state-wise tracking, scenario history, Allure links, search and controlled owner reassignment.
• Creates Jira defects grouped by failure, by owner and state, or by owner, with a dry-run mode to check before filing.

Final-year project

Hope: emotionally intelligent robotic companion

A Raspberry Pi companion robot that reads emotion from face, voice and text, and responds with empathy.

• Multimodal emotion recognition with DeepFace (face), Wav2Vec2 (speech) and Transformers (text).
• Empathetic conversation through Gemini, with memory across sessions.
• Real-time vitals from a BLE smartwatch.

Hotel Management System

Full-stack hotel booking platform with room search, bookings and an admin dashboard.

• Room availability checks and date-range validation for bookings.
• Admin dashboard to add, edit and delete rooms and manage bookings.
• JWT authentication with Spring Security, room images stored in AWS S3.

BMC Hackademia top 3 finalist · 48-hour hackathon

RAG document Q&A

Upload a PDF and ask questions about it.

• Upload one or more PDFs. They are split into chunks, embedded with Hugging Face all-MiniLM-L6-v2 and indexed in FAISS.
• Questions are answered by an LLM through Groq, using only the uploaded documents. Unrelated questions get a “not related to the uploaded documents” reply instead of a guess.

My part:

Built end to end: the Flask backend, the LangChain retrieval pipeline and the web UI.

Personal Finance Management System

Desktop app to track income, expenses and savings goals.

• Normalized (3NF) MySQL schema with triggers, accessed over JDBC.
• Transactions, savings goals with monthly progress, and financial summaries.
• Swing dashboard with XChart charts of monthly income and spending.