Spriha Pandey

Led engineering from prototype to production, building the core product using Next.js 14 (TypeScript) frontend, FastAPI (Python) backend, Docker/Compose for containerized dev, and PostgreSQL for persistence. Working directly with customers to identify bottlenecks in the real estate workflow and engineer solutions to meet those needs Designed and implemented the app architecture—Next.js + Tailwind/shadcn UI, FastAPI + SQLAlchemy, async DB (asyncpg), Redis + Celery workers, and MinIO for S3-style asset storage—to ensure scalability and operability.

Founding Engineer at Eukreate, Inc. (2026-06 – Present)

• Led engineering from prototype to production, building the core product using Next.js 14 (TypeScript) frontend, FastAPI (Python) backend, Docker/Compose for containerized dev, and PostgreSQL for persistence.
• Working directly with customers to identify bottlenecks in the real estate workflow and engineer solutions to meet those needs
• Designed and implemented the app architecture—Next.js + Tailwind/shadcn UI, FastAPI + SQLAlchemy, async DB (asyncpg), Redis + Celery workers, and MinIO for S3-style asset storage—to ensure scalability and operability.
• Built and shipped end-to-end features across the full stack: UI components, API routes, background jobs, DB schemas/migrations, and tests (pytest).
• Collaborated with founders on product strategy, technical roadmap, and feature prioritization to align engineering work with business goals.
• Defined early engineering workflows: Git PR conventions, TypeScript/Python linting and typing, automated tests, and Docker-based local/dev deployment processes.
• Rapidly prototyped and iterated on product ideas across Next.js, FastAPI, and worker pipelines to validate UX and product-market fit.
• Partnered on UX and product design to deliver feasible, polished interactions using Tailwind and shared component libraries.
• Built a scalable foundation for growth: robust Postgres schemas, async query patterns, background processing, TTS integration (ElevenLabs) for media workflows, and containerized infra.

Caliber Research Group (2026-06 – Present)

• Designing and evaluating LLM-powered agent workflows for educational applications including quiz generation, concept explanation, and FAQ assistance.
• Engineering prompts and evaluation pipelines to improve the quality, accuracy, and usefulness of AI-generated educational content.
• Developing automated systems for question generation, answer validation, and grading support using large language models.
• Conducting experiments on AI safety, fairness, and educational effectiveness in programming education environments.
• Analyzing model performance and user feedback to iteratively improve AI-assisted learning tools.

3D Reconstruction Project at UCSB (2025-09 – 2026-03)

• Built an end-to-end 3D reconstruction pipeline using COLMAP, PyTorch, and Octree-GS to generate neural scene representations from multi-view image datasets.
• Developed preprocessing pipelines using Segment Anything Model (SAM) to process complex, unstructured visual data.
• Improved reconstruction quality by implementing mask-guided point cloud densification techniques.
• Debugged CUDA, PyTorch, and numerical compatibility issues within a large research codebase.
• Collaborated with a multidisciplinary research team to deliver machine learning solutions on schedule

Bachelor of Science (B.S) in Computer Science – University of California, Santa Barbara (UCSB) (2026-09)