U.S. patent application filed
Named inventor, with a stealth AI startup. U.S. Application No. 19/783,601, patent pending under Track One prioritized examination.
I'm Devavrath Sandeep, an AI engineer finishing an M.S. in AI Engineering – Biomedical Engineering (AIE-BME) at Carnegie Mellon. I build agentic LLM systems and clinical machine learning that is calibrated, audited for shortcuts, and tested on data it has never seen.
AI ENGINEERING, GROUNDED IN BIOLOGY.
Named inventor, with a stealth AI startup. U.S. Application No. 19/783,601, patent pending under Track One prioritized examination.
Second author at IEEE ICHI 2026 on a digital process twin of breast pathology workflows, plus a paper on desert bird nests in Ornithology Research (vol. 34).
Representing the biomedical engineering graduate program to prospective and incoming students.
Self-evaluating agent pipelines with human-in-the-loop guardrails, a Jacobian-based interpretability lens on Gemma, and deployment from Jetson edge devices to autoscaling GPUs on Google Cloud.
Instruction support, grading, and office hours for the introductory biomedical engineering course.
Proctoring exams for students with accommodations, keeping every sitting fair and secure.
Real-time MediaPipe/OpenCV hand kinematics and a joint-angle-to-actuator mapping for closed-loop exoskeleton calibration.
GPA 3.91/4.00. Deep learning, machine learning for AI engineers, AI systems toolchains, and biomedical AI.
Built a bioinformatics pipeline on HPC clusters to estimate polygenic risk for Parkinson's in Indian cohorts from GWAS data.
Stem-cell phenotyping by flow cytometry (CD34/CD45), HLA typing for transplant compatibility, and apheresis sample processing.
Ran assignments, quizzes, and grading for a class of 236 undergraduates, keeping feedback on time.
Graduated in the top 1% of the department (GPA 7.97/10), specializing in the Science of Intelligence: neuroscience and brain-computer interfaces.
I'm always open to discussing new projects, research collaborations, or opportunities.