Siddh Patel

Groq Llama-3.1 · FAISS · MiniLM · Flask · Next.js · Docling · Docker

• Architected a production RAG pipeline — MiniLM embeddings over a 2,908-chunk FAISS index, Llama-3.1 generation on Groq — deployed live with 100+ active users and median query response under 2 seconds.
• Engineered a bilingual NLP pipeline (Hindi/Hinglish) with mood tracking and crisis detection; processed 500+ documents via Docling and Mistral OCR into a structured knowledge base.
• Presented to Parul University’s wellness cell as a scalable first-layer support tool for 10,000+ enrolled students, complementing existing counseling infrastructure.

PyTorch · FastAPI · WebRTC · Celery · PostgreSQL · Redis · Docker

• Developed a PyTorch computer-vision model for real-time decimal-precision scoring of 10m Air Rifle and Pistol competitions via WebRTC, used by a shooting academy with 40–45 daily athletes.
• Designed PostgreSQL and NoSQL schemas for athlete performance analytics with Redis caching (sub-100ms responses); integrated Chitraksha’s mental-conditioning engine for pre-competition readiness feedback.

Django 5.2 · PostgreSQL 18 · Redis · Celery · Docker

• Built a full-stack Django enterprise system with role-based access control across 4 roles (Admin, HR, Manager, Employee), supporting bulk CSV and Excel attendance upload with server-side validation.
• Implemented asynchronous Celery and Redis email notification pipelines; containerized the full stack with Docker for consistent production deployment.

• Engineered a Django-based CRM and HR system managing 1,500+ employee records with PostgreSQL, role-based access control, and REST APIs for bulk CSV attendance upload.
• Optimized server-side performance with Celery and Redis async email notifications and background task processing.
• Containerized the full deployment stack with Docker, reducing environment setup time by 60% and improving release consistency across development and production.