❯ ls ./projects
// Things built, shipped, and learned from
AI-powered supply chain intelligence platform built as a Master's final project. ML models for delay and cost prediction, an NLP chatbot for real-time data queries, and automated news monitoring using BERT and OpenAI. Fully orchestrated with Airflow and deployed on AWS.
Multi-tenant SaaS for marketing agencies. Dual Node.js backend separating dashboard concerns from lead ingestion, per-client RBAC, custom lead fields, and server-side Meta Conversions API firing on lead conversion. Deployed to production on AWS EC2 with RDS, MongoDB Atlas, and CloudWatch. 157 automated tests across 19 suites. Private client project.
From requirements gathering to production. Identified Odoo as the right open source ERP for a client, then deployed and hardened the full stack: EC2 with Nginx reverse proxy, RDS, S3 file storage with Glacier lifecycle, Let's Encrypt SSL, CloudWatch alarms, WireGuard VPN for remote access, and office IP restriction at the Nginx level. Private client project.
Natural language interface for your database. Connect it to any SQL or NoSQL source and ask questions in plain language. The agent translates intent into queries, runs them, and returns structured insights, trend summaries, and decision-ready answers. No SQL knowledge required on the user end.
An autonomous AI assistant that handles the overhead of being reachable. It reads your emails and calendar, answers incoming calls, transcribes and summarizes what was said, books meetings, drafts replies, and pushes you a clean daily brief. The goal: zero inbox anxiety, zero missed context.
A console game written in x86 assembly. No frameworks, no abstractions, no shortcuts. Just registers, memory addresses, and a genuine understanding of how machines work. Built before anyone could ask an AI to do it for them.
❯ ./collaborate.sh
Open to projects and collaborations.
./get_in_touch