Sanan Malik sananmalik

class Sanan:
    university = "MNS University of Agriculture, Multan"
    degree     = "BS Computer Science · '28"
    stack      = ["C#", "ASP.NET Core", "SQL Server", "EF Core"]
    ai_tools   = ["OpenAI API", "Gemini API", "Multi-Agent Systems"]
    ml_active  = ["Python", "NumPy", "PyTorch (learning)"]
    building   = "AI-integrated backend systems → research"

⚡ Current Focus

🔧  Building production backend systems in C# / ASP.NET Core
🤖  Wiring AI inference into real pipelines — not just calling APIs
🧠  Learning ML foundations to understand what's inside the models I deploy
🎯  Targeting -

🚀 Projects

🔍 SentinelIQ

Web Intelligence Platform

Real-time competitor monitoring — scrapes live sources, runs OpenAI inference, returns structured business signals through a layered REST API.

Confidence-gated output filtering prevents hallucination drift from compounding silently across pipeline stages.

🛡️ SIFT Guardian

Multi-Agent Security Forensics · UC Berkeley AI Hackathon

Four specialized agents — Investigator, Skeptic, Verifier, Reporter — running a self-correction loop over security telemetry.

Inter-agent confidence scoring lets the Skeptic halt and redirect the Investigator mid-analysis before bad data reaches the final report.

🧠 Cognivex AI

Academic Workflow Agent · Live on Vultr VPS

Parses academic documents across formats, extracts tasks and deadlines, generates structured study plans. Deployed behind Nginx.

Extraction confidence checks surface incomplete parses explicitly — the agent admits when it cannot plan rather than guessing over bad data.

🎓 STEMPilot AI

Adaptive Learning Platform · DSH Hacks V1

Identifies student knowledge gaps through quizzes, tracks performance, and adjusts content delivery based on results.

Co-built under hackathon constraints — focus was adaptive scoring logic and real-time feedback loops.

🛠️ Tech Stack

📌 What I'm Building Toward

The research question I keep running into across all these projects:

How do you build agentic systems that know what they don't know — and surface that uncertainty reliably instead of hallucinating confidence?

Open to research collaboration and backend/AI internship opportunities

Navigation Menu