laksh jain

I’m an early-career machine learning engineer focused on turning advanced modeling into dependable products—especially in safety, security, and decision support. I enjoy building systems that are measurable, explainable, and resilient in real deployment conditions.

As a Machine Learning Intern with IBM Cloud SkillBuild (Remote, AICTE), I deployed a Machine Health Monitoring dashboard that ingests sensor fault data and triggers predictive maintenance alerts through an ML backend pipeline. I tuned a Random Forest model to an F1-score of 0.89 (92% critical failure capture, <5% false alerts), configured an AutoML workflow in Watson Studio, and deployed the selected model to IBM Cloud with dashboard integration.

My projects blend neuro-symbolic modeling, retrieval, and security: I built an A3TGCN2-based Smart-Home Safety Framework over 1M+ telemetry records, added a Physics Guardrail to eliminate sensor-induced hallucinations, and achieved 92% accuracy across forensic scenarios. I also developed PolicyAI (real-time RAG with cited verdicts and a claim feasibility scoring workflow) and Jjawan, a zero-trust defense system with dual-pipeline message security for LLM-based threat analysis—earning Runner-up at the Smart India Hackathon (SIH) 2025.