About
I'm an AI Engineer with hands-on experience building production-ready AI systems that turn large language models into practical, user-facing tools. My work focuses on applied Retrieval-Augmented Generation (RAG), Agentic AI workflows, and scalable backend services that support real-world use cases.
Currently, I'm an AI Developer at Offshorly, where I develop MCP-based agents, real-time AI features, and workflow-driven integrations for enterprise platforms. Previously at LexMeet, I worked on legal AI systems—implementing RAG pipelines, optimizing inference performance, and building document intelligence tools that reduced processing latency and supported over 100+ active legal practitioners.
Outside of day-to-day development, I enjoy experimenting with computer vision projects (YOLO), improving prompt and retrieval strategies, and building automated pipelines for large-scale data ingestion and evaluation.
Experience
Developing enterprise-grade AI features including Model Context Protocol (MCP) server implementations and automated workflow triggers. Building user-personalized AI responses using CRON and agentic tools for real-time collaboration platforms.
- MCP
- OpenAI
- React
- Node.js
- Laravel
- Pinecone
Reduced inference latency by 45% for the AI Docs platform. Implemented production-grade RAG pipelines using FAISS and built automated scraping pipelines for 60,000+ legal cases. Developed an AI Contract Verifier for intelligent clause extraction.
- Python
- RAG
- FAISS
- Llama 3
- GPT-4o
- RunPod
- FastAPI
Developed 20+ reusable React components and implemented RESTful API endpoints. Automated an image compression pipeline for 2,000+ company assets, significantly improving page load times on GCS.
- React
- Node.js
- MySQL
- Jest
- GCP
Projects
A comprehensive suite of legal AI tools including LexDocs (template retrieval), LexAssist (RAG-powered legal advisor), and AI Contract Verifier. Achieved 14/15 legal accuracy rating in attorney-led testing.
- FAISS
- LangChain
- FastAPI
- React
- LLM Chaining
Multimodal AI system combining YOLOv7 and StrongSORT for real-time human pose detection and classroom engagement tracking. Achieved 0.86 mAP@0.5 on custom precision datasets.
- YOLOv7
- StrongSORT
- PyTorch
- Computer Vision