BlackCloud

Project Concept

You I/ You Ex is the new UI/UX

Entry

Status: Submitted

Last saved: February 21 at 6:28 PM EST

View Entry

Team Roster

You must be registered for the event to view the team message board.

Tanishq Sharma Team Lead RSVP Approved

AI Engineer at JerseySTEM

Frontend & Interactive Physics (The Canvas) Technologies: Next.js 14, @xyflow/react, Framer Motion. Responsibilities: Built the spatial 2D workspace and draggable node interfaces. Implemented the Framer Motion spring physics for the "Semantic Gravity" drifting animations. Developed the dynamic, Claude-generated UI components (sliders, star ratings, vibe meters) that shatter traditional chat UI conventions. Created the playful interactive "Magnet" tool to pull filtered cards across the canvas.

I build deployable AI systems that run fast, reason across modalities, and integrate into real workflows. My work spans low-latency computer vision on NVIDIA Jetson (sub-5 ms inference), RAG pipelines with LlamaIndex and vector databases, multimodal image captioning with EfficientNet + Transformers, and real-time data platforms using Kafka, PySpark, and Streamlit. Currently an AI Engineer, I designed and deployed production RAG chatbots that connect WordPress, MySQL, and Pinecone to deliver grounded, sponsor-aware responses. I am also developing agentic workflows for automated content generation and data retrieval using Gemini and LLM tool-use patterns.

Applied Machine Learning, Computer Vision, Multimodal Solutions, low-latency model deployment, LLM-powered retrieval-augmented generation systems, real-time AI systems

Previously, I built and patented a real-time pothole detection system deployed on Jetson Nano for edge road-safety analytics. Recent projects include: • Multimodal image captioning (EfficientNet-B0 + T5) with custom projection layers • LostSight VLM pipeline for unattended object detection with tracking • Real-time crypto ETL platform (Kafka → Spark → PostgreSQL → LSTM → Streamlit) • Clone screening Monte Carlo simulation dashboard for biotech decision support

Arvind Chary Padala RSVP Approved

Student at Rutgers University

Backend, AI Engine, & Vector State (The Brains) Technologies: Python, FastAPI, Claude 3.5 Sonnet, Redis Stack (JSON + VSS). Responsibilities: Engineered the stateless backend architecture using lightweight FastAPI wrappers to ensure stability. Integrated Claude 3.5 Sonnet to drive the generative UI schemas from unstructured inputs. Implemented Redis JSON to persistently store card state natively alongside their Float32 embeddings. Leveraged Redis Vector Similarity Search (VSS) to calculate lightning-fast, real-time cosine similarity, powering the underlying math for the semantic gravity feature without secondary database lookups.

I'm Arvind Padala, a data scientist studying for a master's degree in data science at Rutgers. I was once a software engineer at Infosys. I enjoy creating AI that can reason in addition to reacting. Creating functional prototypes from complex concepts is something I enjoy doing, whether it's utilizing LangGraph and CrewAI to develop agentic AI systems or training transformer models for movie recommendations. My focus is on developing intelligent systems that address real-world issues, and I have over two years of expertise with automation, analytics, and AI pipelines. My experiments are also shared on www.arvindchary.io.

Agentic AI, Autonomous Systems, Gen AI, Casual Inference in AI, Deploying Pipelines