ML Engineer @Medical Guardian — teaching machines to predict fall risks in elderly patients. Yes, really. No, it's not boring.
class SalmanKarim:
def __init__(self):
self.role = "ML Engineer"
self.current_mission = "Fall risk prediction for PERS devices. Clinical data → decisions that matter."
self.stack = {
"ml_core": ["XGBoost", "Scikit-learn", "MCDA", "KNN", "Drift Detection"],
"data_pipeline": ["dbt", "SQL Server", "Medallion Architecture", "Python"],
"ai_brain": ["LangChain", "LangGraph", "RAG", "FastAPI", "n8n"],
"frontend_magic": ["Next.js", "React", "React Native", "TailwindCSS"],
"cloud_ninja": ["AWS", "Cloudflare", "Vercel", "DigitalOcean"]
}
def daily_routine(self):
return [
"☕ Coffee",
"🔍 Audit clinical data",
"🤖 Train & threshold-tune ML models",
"💻 Ship production pipelines",
"🔥 Break something",
"🛠️ Fix it (mostly)",
"🎮 Repeat"
]🏥 Medical Guardian — ML Engineer (Feb 2026 – Now)
Building a 30-day fall risk prediction system with XGBoost, patient-isolated splits, and recall-optimized thresholds. End-to-end Medallion Architecture on SQL Server. Healthcare data quality, drift monitoring, and model reliability in a high-stakes production environment.
🏛️ STZA — AI & Digital Solutions Engineer (Oct 2025 – Feb 2026)
Automated government workflows with LangChain, RAG, and n8n. Made bureaucracy faster. Somehow survived.
🚀 Beyond Future Group — Software Engineer (Jun 2024 – Sep 2025)
Cloud-native e-commerce, AWS deployments, CI/CD pipelines. Mentored juniors who now know more React than me.
📱 AXXON Solutions — Full-Stack Mobile Developer (Jun 2023 – Jul 2024)
Shipped React Native app with AI chatbots and real-time booking. Survived production bugs.
The ML/AI Core
The Frontend Magic
The Cloud Game
| Project | What It Does | Stack |
|---|---|---|
| Medical Guardian ML | Fall risk prediction · 96.2% recall · 54% ↓ operator calls | XGBoost · FastAPI · Scikit-learn |
| Travel Planner Agent | Multi-agent Budget + Research + Itinerary orchestration | LangGraph · FastAPI · SQLite |
| Hunza RAG Advisor | AI travel guide with strict source grounding | Python · RAG · Embeddings |
| Gulmit.info | Tourism platform with ISR + AI content pipelines | Next.js · Firebase · Cloudflare |
📬 salmankarim.is-a.dev · Open to collabs on ML, AI, or anything with real-world consequence 🚀
🎮 Fun Facts About Me
- I once debugged a model for 6 hours. The bug was a data leakage in the train split.
- I build tourism platforms for my hometown in Hunza because why not
- 18 years of Boy Scouts — yes, that's on my CV
- Coffee-to-code ratio: 1:1000 lines
- Night owl who commits at 3 AM (sorry, teammates)
- When the model diverges, I go play guitar