A collection of enterprise AI solutions, agentic systems, and production deployments showcasing real-world impact across fintech, healthcare, and design innovation.
Multi-agent AI system for Samsung's wearable technology design optimization. Combines generative AI, 3D geometric processing, and predictive ML models for autonomous design assistance.
Enterprise RAG system for Samsung Design Innovation Center with advanced retrieval strategies, semantic chunking, and hybrid search for design documentation and research materials.
AI-powered KYC verification system for regulated fintech startup with GDPR compliance framework, document verification pipeline, and comprehensive audit trail capabilities.
Enterprise RAG system for startup evaluation across multiple innovation hubs. Multi-stage retrieval pipeline with query optimization and reranking, serving 500+ requests/hour.
Credit risk assessment platform for National Commercial Bank Jamaica with ML-based scoring models, explainability features, and Caribbean banking regulations compliance.
Production-grade agentic AI system for automated financial analysis in regulated fintech environment. Multi-agent architecture with specialized agents for secure data retrieval and compliance checks.
LLM-powered backend for generating personas from unstructured research. Document processing, ChromaDB, multi-step persona generation, and prompt completion.
Upload, parse, and download a polished CV using OpenAI. Vite + Express app for turning messy CV text into clean, structured output.
Cost-aware, fair resume screening using Bayesian orchestration over multiple LLMs. Reduces hiring costs and improves demographic parity.
Modern Goodreads alternative for research papers: browse, track, rate, and get personalized recommendations with TF-IDF and sentence-transformers.
Voice-powered AI agents for conversational interfaces and voice-driven workflows—STT, agent/LLM, TTS in a modular pipeline.
Generate 3D objects from text and images with VLM-enhanced prompts; Shap-E, Hugging Face, Neural4D, Instant3D; export OBJ, PLY, STL.
Multiple LLMs deliberate together to answer hard questions—completion, judging, and consolidation for better robustness.
Experiment with prompts, compare models, and iterate on LLM inputs and outputs in one place.
LLM-based support with LangGraph: RAG, NLP-to-SQL, and analyzer agents with automatic query routing. FastAPI, Pinecone, React.
Estimation and forecasting for margins, effort, or similar quantities—configurable inputs and logic, repeatable outputs.
End-to-end observability demo with synthetic microservices, Prometheus metrics, anomaly detection service, and Grafana dashboards. Features real-time monitoring and early-stage AIOps capabilities with live GitHub Pages demo.
Modular AIOps toolkit for log analysis with template extraction, semantic anomaly detection, error burst detection, and pattern drift detection. Features TF-IDF vectorization and Isolation Forest for intelligent log monitoring.