Hi, I’m Karthik.
CS + ECSE student at RPI building applied AI/ML systems and shipping production software. I like projects where data pipelines, measurable evaluation, and real deployment constraints matter.
Now
- Building Loxe AI (SOC 2 automation + AI security analyst).
- Dynamic ASL recognition (MediaPipe + transformer sequence model).
- Seeking Summer 2026 ML/SWE/AI infrastructure internships.
Highlights
- Applied for pre-seed incubator for Loxe.
- Hybrid classifier: ~85% accuracy; loss ~0.61 → ~0.36 (best ~0.15).
- Emotion recognition: ~70% validation on 30k images.
Stack
PythonTensorFlow/KerasMediaPipeTypeScriptNext.js/ReactPostgreSQL/PrismaAWSTerraform
Featured work
Loxe AI — SOC 2 Automation + AI Security Analyst
B2B SaaS platform that scans AWS posture and streams Terraform remediation via an AI analyst.
Next.jsTypeScriptAWSTerraform
Dynamic ASL Recognition (MediaPipe + Transformer)
Landmarks → 30-frame kinematic tensors → time-series transformer for dynamic gesture classification.
MediaPipeTensorFlowTransformers
Hybrid Quantum-Classical Image Classifier
Optimization-driven training (genetic algorithm) with stable convergence and loss reduction.
OptimizationPythonML
Facial Emotion Recognition (Assistive Tech)
CNN/MLP emotion recognition with a focus on evaluation and generalization.
Computer VisionCNNsTensorFlow