Made live demos more realistic by improving browser behaviour and flow fidelity, creating a closer-to-product experience for prospects. The work is live in production.
Building what works.
Two internships. Production systems, applied AI, and practical engineering decisions.
Omnisavant.ai — Software Development Engineer Intern
Working on production systems behind interactive product demos — improving how realistic the demos feel, how quickly mock environments can be created, and how product flows are turned into reproducible browser automation.
Built a programmatic sandbox-cloning and Claude-assisted mock-generation workflow that turns cloned product environments into usable mocks in roughly 3–6 hours. The workflow is live in production.
Contributed to the production demo orchestration pipeline, including demo-plan generation, Playwright script generation, workflow execution and validation.
amasQIS.ai — Artificial Intelligence Engineer Intern
Built backend and applied-AI features across two kinds of products: industrial analytics and an AI-powered virtual try-on experience. My work spanned data access, agent tooling, REST APIs, reporting, and model integration.
SQL-connected industrial assistant — Implemented data-retrieval tools using FastMCP and LangChain, integrating LLaMA 4 Scout to help analyze electrolyzer measurements and answer questions grounded in industrial data.
Automated technical reporting — Built diagnostic reporting flows that turned SQL results into visualizations and structured Quarto/PDF outputs, making technical analysis easier to review and share.
FastAPI backend for virtual try-on — Worked on authentication, user profiles, body measurements, settings, community posts and AI prompt/search endpoints; integrated Firebase services alongside Gemini, MediaPipe and OpenCV for pose-aware experiences.
Predictive-model and delivery pipelines — Worked with LSTM and XGBoost modeling, PyTorch-based experiments, FastAPI serving, Docker packaging and Airflow-based workflow orchestration.