Portfolio 2026
Divya
Tripathi
Backend Systems · AI/ML Engineer · RAG Systems
Hi, I'm
Software Engineer
Divya
Tripathi
Curious enough to question, ambitious enough to build.
Available for internships & full-time opportunities.
Based in Delhi, India
Available Worldwide
Divya
Who I Am
Software Engineer with experience in backend systems and AI/ML applications, specializing in building REST APIs, retrieval-augmented generation (RAG) systems, and scalable services using Python, FastAPI, Docker, and PostgreSQL. Strong foundation in Data Structures & Algorithms, DBMS, Operating Systems, Computer Networks, and System Design, with hands-on experience across LangChain, LangGraph, vector search, and multimodal machine learning.
Vizh AI Solutions
AI Engineer Intern
- Developed a production-ready Retrieval-Augmented Generation (RAG) chatbot for course-based knowledge retrieval using LangChain, FastAPI, and PostgreSQL (pgvector), enabling context-aware question answering.
- Designed LangChain agent workflows for query routing, context retrieval, and answer validation, improving response quality across AI-powered learning workflows.
- Built scalable FastAPI backend services with asynchronous processing, semantic search, and caching, improving end-to-end response latency by ~40%.
- Containerized backend services using Docker Compose and integrated PostgreSQL with Prometheus monitoring, implementing persistent conversational memory using PostgreSQL (JSONB) for personalized responses across multiple user sessions.
Delhi Technological University (DTU)
Research Intern
- Developed a multimodal emotion-recognition system (text + speech), achieving 86.8% text and 81.2% speech classification accuracy.
- Preprocessed and trained ensemble deep learning models on 18K text and 6K speech samples.
- Engineered an inference pipeline and deployed the model for real-time emotion classification across multimodal inputs.
QuickGhy
AI Engineer Intern
- Developed a content-based recommendation engine for global exporters using TF-IDF vectorization and cosine similarity to identify relevant international trade opportunities.
- Developed an automated data collection pipeline extracting 3K+ records using Selenium and BeautifulSoup.
- Presented the recommendation system at the Global TradeTech Forum (World Economic Forum).
