1
1
2+ years of experience
AI & ML Engineer specializing in production-ready AI systems — Computer Vision, RAG architectures, and AI Agents. Built and shipped: - BistroFlow AI — live WhatsApp SaaS for Pakistani restaurants (Roman Urdu NLP + real-time dashboard) - DeepFakeShield — 95.67% accuracy deepfake detection system (FYP, sole developer) - EduMentor Live — multimodal AI tutor deployed on Google Cloud (Gemini Hackathon) - ExportShield Pro — commercial offline desktop app for ZATCA compliance (shipped .exe) Hackathon builder: Google Gemini Live API · GitLab AI 2026 Stack: PyTorch · FastAPI · React · Supabase · n8n · Pinecone · Claude API · Gemini API Open to Junior AI Engineer roles — Remote & Lahore

Project Aura is a high-performance digital forensic intelligence platform designed to bridge the gap between complex cyberattack data and actionable insights. Built specifically for the AMD Global Hackathon, the system is powered by the AMD Instinct™ MI300X accelerator, utilizing the ROCm™ 6.0 software stack and vLLM for ultra-fast neural inference. The core innovation lies in its "Neural Forensic Pipeline," which takes raw system logs, network captures, and MFT (Master File Table) entries to reconstruct attack timelines automatically. By leveraging the Qwen-7B-Instruct model, Aura can identify stealthy 'Living off the Land' (LotL) techniques, C2 beaconing patterns, and credential harvesting attempts that traditional tools might overlook. Key features include: AMD Optimized Inference: Direct integration with MI300X for real-time analysis. Automated Report Generation: Converts AI findings into professional, court-ready PDF documents. Minimalist Interface: A streamlined Streamlit dashboard for investigators to perform rapid audits. Project Aura demonstrates how cutting-edge AMD hardware can revolutionize cybersecurity by automating the most time-consuming parts of digital forensics.
10 May 2026