Multi-disciplinary senior specialist support for signal processing, algorithms, In Vitro and In Vivo data collection, and procedure design. In addition to design and implementation, I provide team guidance and with them solve the "worrisome" technical problems that are stalling your product roadmap.
Your engineering team is likely stretched. The prototype has noise that is slowing or blocking algorithm development. You need training, testing, and validation data but you don't have the experience to make sure the data collected will do the job. This might even be your second attempt to obtain data.
These aren't just technical glitches. They are compound interest on your project timeline. Every week of delay burns runway and credibility.
Many times I've navigated the messy path from concept to product. I use that experience to identify & solve problems, and guide your team away from dead ends. I use AI LLM methods to come up to speed quickly on problems and to understand existing Gold Standard approaches and improve upon them, and benchmark new solutions. I work with your team of physicists, engineers, and computer scientists, to architect robust systems. I don't just advise; We learn together and we execute.
Handling execution details so you can focus on high-value decisions. Increasing bandwidth by working across disciplines to meet company objectives.
Noise reduction, filtering, and optimizing signal chains for hardware prototypes and later revisions.
Developing, training, testing, and validating algorithms to extract actionable metrics from complex datasets.
Designing In Vitro & In Vivo protocols. As needed, I work directly on site, in the clinic, or OR, to ensure data gathered are exactly what the algorithm team needs.
I am not a distant freelancer. I operate as a plug-and-play senior team member, removing the administrative and logistical friction of hiring.
Hire senior expertise while maximizing your budget. As a Canadian resident operating through my Canadian Corporation I offer significant financial leverage for International (ie, US & European) entities.
Select a project to view the Problem, Action, and Impact.
EEG-data based University prototypes for detecting markers of Clinical Delirium worked in ideal conditions, but were not robust to be successful in hospital environments. Further, the existing university studies lacked the datasets and analysis needed to meet CE and FDA approval requirements.
Senior Algorithm Engineer and the 3rd person joining the company: to "translate" a university prototype technology so that it would work in real-world settings and to generate the robust performance necessary for FDA/CE approval.
I transitioned the technology from Fourier-based analysis to a Wavelet architecture to facilitate classification and artifact removal methods. I created a strategy to detect garbage data (no-decision) when present. I developed ways to, in real-time, estimate how much additional data are needed to provide classification output. I provided the team with necessary clinical subgroup and performance data to achieve CE and FDA approvals.
Achieved FDA (USA) and CE (Europe) approval for a commercial hospital product. The product minimised "Decision Energy" for clinical staff by providing automated Clinical Delirium identification to an accuracy that is statistically indistinguishable from the current Gold Standard.
Identifying and tracking arterial vessel walls in real-time ultrasound data is hindered by tissue motion and low Signal-to-Noise Ratios (SNR), making automated blood pressure estimation unreliable. Further, dynamics of vessel rigidity add challenges to creation of a reliable model for blood pressure.
Senior Algorithm Engineer and 4th person to join the company: to create the foundational IP for automated blood pressure monitoring using ultrasound and validate the physics model through In Vivo animal studies. To create an automated way to detect the location of arterial vessel walls, measure their movement, and estimate blood pressure from ultrasound data.
I established the IP base by creating a 100% automated algorithm for vessel wall identification and tracking in M-mode data. Using their previously collected In Vitro and In Vivo data, I proposed and implemented an initial algorithm to automatically find and track vessel wall movements. I created a data processing pipeline suitable for statistically comparing the performance of algorithm changes and dataset variations, for both In Vitro and In Vivo data. Using an empirical model, I demonstrated how error in the measured quantity relates to error in the desired estimated value, blood pressure. In preparation for collecting In Vivo Ovine data, I with the team and the OR Veterinarian and staff created a robust data collection protocol, and accounted for challenges collecting data in the OR. I additionally joined the team in the OR to ensure the quality and viability of collected data.
Positioned the company for rapid algorithm iteration by delivering a custom data processing pipeline and validated Ovine data within 6 months. Created a strategic IP moat around automated vascular health quantification.
Reliable acquisition of clinical-grade vitals (like Potassium, SpO2, and Heart Rate) from a single wearable patch on ambulatory dialysis patients is historically compromised by motion artifacts and low signal fidelity.
Senior Algorithm Engineer: brought onto the team to facilitate communication between the "deeply talented" ML team and the product team. Additionally, to provide assistance with strategy development, determining approach feasibility, prioritizing work, and DSP assistance.
I led weekly progress presentations helping the technical team communicate their accomplishments/challenges and helped the business and technical teams coordinate next steps and IP development strategies. I created wavelet-based methods to process acoustic sensor data to isolate features of an ECG-like signal obtained over large vessels with the goal of parameterizing these features to determine vessel health. I assisted the team in obtaining more useful PPG features by providing a baseline removal algorithm (outperforming Butterworth filters or other standard methods) suitable to remove very large momentary events and baseline wonder without compromising target event characteristics. My knowledge and methods augmented purely machine-learning based methods to isolate the desired signals. I created and implemented adaptive filtering techniques for feature extraction in non-stationary noise and signal environments.
The interaction among the teams shifted to a "learn together in baby-steps approach" and collaboration increased. Introducing DSP methods to the "highly technical ML team" reduced ambiguity with which they were faced, gave them extra tools, and could more easily determine the impact of their algorithms.
The team faced technical hurdles in isolating photonic emissions related to mitochondrial oxygen uptake from multiple interfering noise classes, stalling performance milestones.
Senior Signal Processing/Algorithm Specialist and 5th person to join the company: to unblock the engineering physics team by solving complex signal-to-noise ratio issues and noise-source unmixing that were stalling feasibility demonstrations.
I implemented methods to address various noise classes and unmix data to isolate target photonic signals. I guided the use of In Vitro/In Vivo data collection and repeated measure statistics to create "tight-cycle" DSP and algorithm comparisons instead of 1-off measure-analysis iterations.
The company demonstrated product feasibility roughly a month after I joined the team. They now possess a validated v1 algorithm and an objective process for comparing signal processing and algorithm iterations to drive engineering decisions.
The firm required a robust tele-medicine prototype capable of high-fidelity scalp EEG acquisition and remote data acquisition/patient behaviour monitoring to facilitate a geographically distributed patient assessment and intervention team.
Technical Lead & Systems Architect and 4th person to join the company: to assist with project road-map and leverage GNU sources to design and build a low-latency data transport architecture required for real-time remote biofeedback. To hire team members as needed to complete the job required.
I partnered with Metiris and established a specialized Canadian team to develop software for seamless EEG hardware integration. We created a data transport architecture designed to move biosignals across LANs and the internet with the low-latency required for real-time biofeedback and remote patient monitoring. When hardware challenges were encountered, I used my prior electronics technologist training to trace the problem and replace components of the custom EEG hardware
The company successfully proved product feasibility and demonstrated a fully functional working prototype. This established the foundation for the firm's distributed health product strategy.
High rates of surgical site infections post-Cesarean section required a low-friction, high-velocity reporting tool to prevent patient complications.
Technical Product Owner working with 2 teams of 6 developers each (joined a company of 50+ people): to translate customer requirements into technical specifications and prioritize the daily work and product roadmap to achieve a compliant MVP.
In my early days with the company, I found that a barrier to success was communication across the company. I resolved this by using the whiteboards throughout the organization to communicate the work underway. When key milestones were met, I spoke in front of the company in my role to communicate the successes and path forward. To fulfill my role, I developed positive peer relationships with the technical teams (together discussing problems and strategies) and developed collegial relationships with the business team.
Successfully achieved a Minimum Viable Product (MVP) and began plans to deploy it to partners who participated in its development.
How to increase decision-making bandwidth and technical expertise at relatively low cost?
Senior R&D Consultant & Mentor hired to act as a "Force Multiplier" providing multi-disciplinary leadership and decision making, and senior technical domain-specific expertise in DSP, ML, and Product Design.
I led technology development meetings and faciliated communication between the highly technical team and external stakeholders, I provided new approaches to team and stakeholder interaction. When needed I have provided technical expertise and perspectives. I guided technical strategy decision making.
Freeing other specialists in the company to work on other projects and priorities while continuing and deepening the relationship among the internal technical team and external stakeholders; advancing the IP portfolio and disseminated skills and expertise.
I have availability for one new contract in Q1 2026.
Book a Consult CallDirect Email: pzeman@clinezeman.com