PhD Bioengineering · Pitt · Cardiovascular Health Tech Lab
I do early-stage investigation on bringing invasive information into the non-invasive world. First chapters: cardiac output and venous pressure. Defending December 2026 — open for what's next.
I build non-invasive hemodynamic monitors — hardware, signal pipelines, ML — that hold up against the catheters they're trying to replace.
I'm a PhD candidate in Bioengineering at the University of Pittsburgh (Biosignals track), advised by Prof. Ramakrishna Mukkamala in the Cardiovascular Health Tech Lab. My work spans custom hardware, signal processing, and machine learning — non-invasive sensors across the whole hemodynamic stack, validated against invasive references in multi-site surgical studies.
I came to Pitt after a Master's in Integrated Circuits from University of Tehran and a Bachelor's in Digital Electronics from Amirkabir University of Technology. Between degrees I spent three years at Tosan Co. leading 14-layer mixed-signal PCB design and microprocessor signal-processing chains — I came to research with a production mindset.
I care about the parts of medical devices most people never see: the PID loops, the noise floor, the calibration drift, the pneumatic transients, the DSP and ML pipelines that turn raw physiology into meaningful vitals. That's where patient safety lives — and where honest accuracy begins.
Cardiac output, blood pressure, HRV, and respiration — pulled out of the body without a single needle. PPG at the fingertip, BP at the arm, ECG on the chest, respiration from the cuff. One patient, many modalities.
Custom PCBs (up to 14 layers), analog front-ends for PPG / ECG / pressure / respiration, pneumatic pump & valve control with PID, bring-up and failure analysis.
PPG and ECG denoising, harmonic-SNR quality gating, R-peak detection, pulse-pressure envelope fits, respiration extraction, and ML for hemodynamic parameter estimation.
An arm cuff doesn't measure pressure directly — it reads volume oscillations from the artery underneath, mapped through a sigmoidal volume–transmural-pressure curve. Drag the cuff pressure: invasive arterial pressure rolls in along the X-axis, gets mapped through the curve, and rolls out as volume on the Y-axis. The output is loudest when the cuff sits exactly at MAP.
I look for clinically grounded R&D teams and early-stage investigation groups — groups shipping non-invasive vitals, wearables, or hospital-grade home monitors.
Start a conversation → Download full CV (PDF)Chased a decaying sine wave off a SAW sensor in 2018. Chasing an oscillometric envelope off an arm cuff now. The physics is different — the tricks are the same. Watch one morph into the other:
An end-to-end non-invasive platform that turns an automatic arm cuff into a hospital-grade hemodynamic monitor — pneumatics, ECG, respiration, and pressure on one system. Intended use: intraoperative and ICU hemodynamic monitoring; an investigational device today, no FDA submission yet.
Smart Cuff integrates custom mixed-signal PCBs (pump/valve PID control, pressure, and ECG front-ends) with a PyQt6 clinical GUI for real-time acquisition, automated calibration, and data collection in multi-site surgical studies. My first-author work on cardiac output estimation from Smart Cuff data earned the Outstanding Paper Award at MACSOS 2025 and is under review (medRxiv 2025).
For clinicians: A continuous, catheter-free trend signal you could trust to direct a fluid bolus or vasoactive titration is the goal — comparable in trending performance to invasive pulse-contour today (concordance 83% vs. its 81%), but without the arterial line. Not yet a replacement for thermodilution at the population level; an addition to the toolkit when an arterial catheter isn't appropriate.
Where it sits: noninvasive cardiac-output landscape includes NICOM (bioreactance), ClearSight / CNAP (continuous finger cuff), and intermittent oscillometric devices. Smart Cuff aims to combine the form-factor of a familiar arm cuff with the trending capability typically reserved for finger cuffs — a single device for both BP and CO, with no consumables.
MATLAB/Python pipelines for PPG, ECG, respiration, and oscillometric signals — harmonic SNR quality gating, inter-slot dark subtraction, R-peak gating, envelope fitting, and respiratory modulation extraction.
PyQt GUI running on Raspberry Pi 5 and Windows, supporting four PPG array configs, MCC128 DAQ, PCA9685 PWM, NI-DAQ, dual-camera edge detection, and GitHub-synced deployment to research stations.
Led 14-layer DSP/IF PCB design (EMI/EMC, PDN, thermal budgeting). Built automated RF characterization workflows and directed microprocessor signal-processing chains on Artix-7 and Kintex-7.
Not a formal QMS role — but I've worked adjacent to these standards on clinical studies and can speak the language from day one.
The questions I get asked most often by recruiters and collaborators — answered up front, so we can use our first call for the interesting stuff.
I will defend in late 2026 and am targeting a January – March 2027 start date. For the right role I can consider a part-time or research-collaboration engagement earlier while I wrap up.
All three work. I'm based in Pittsburgh but open to relocating anywhere in the U.S. — East Coast, West Coast, or anywhere with an active medical-device R&D ecosystem. For a great team I'll also consider Europe or Canada.
I do my best work in early-stage investigation groups for R&D — clinically grounded teams shipping non-invasive vitals, wearables, or hospital-grade home monitors. Small-to-mid size ideal; I'll happily own a cross-discipline lane.
My clinical studies are IRB-approved and follow human-subjects research protocols. I haven't led a formal 510(k) or IDE submission, but I've worked adjacent to design controls, IEC 62304 software lifecycle, and ISO 14971 risk management — and I can speak the language from day one.
Both — honestly, I'm an embedded systems engineer at heart, so I code and design electronics together; which side leads depends on the project. I ship the whole stack: custom PCB → firmware → signal processing → ML → clinical GUI. Python daily (Pandas/SciPy/PyTorch/PyQt), C/C++ for embedded, MATLAB for quick prototyping, Git for everything.
Available on request — my PhD advisor and clinical collaborators (PIs and Co-PIs). Reach out by email and I'll share contact details.
Email is fastest: Mahdi.Jazini@pitt.edu. LinkedIn DMs work too. I reply within a day or two.
Open to clinical collaborations, lab visits, or a deep dive on signal processing. Based in Pittsburgh, PA — open to relocation, hybrid, or remote. Best reached by email; I usually reply within a day or two.