Jason Sroka Email & Phone Number

Lunenburg, Ma, Us

Cambridge, Ma, Us

Palm Beach, Florida, Us

Leading Aidéo’s Data Science and Analytics teams, leveraging AI and machine learning advances to transform the medical coding industry, helping to offset the growing coder shortage:• Hired to propel Data Science efforts. Introduced significant staffing changes including refreshing core team and augmenting with Analytics and Reporting team members, establishing tailored agile processes across onshore and offshore teams.• Oversaw modeling enhancements that delivered our first full claim generation automation for medically complete claims, meeting the demanding 95% full claim prediction accuracy as required (and our customers audit!).• Expanded revenue-generating medical coding domains from one (Radiology) to now four (adding Ambulatory Surgical Centers, Anesthesia, and Evaluation and Management), with over twenty-fold increase in volume• Delivered >90% response rates with increasing portion fully automated, including features that go beyond medical coding to meet our customers’ varied needs and preferences.• Developed metrics and reporting demonstrating 50% reduction in manual effort for both external and internal claims coders across multiple medical domains using claim predictions where we couldn’t fully automate.• Worked closely with Revenue Cycle Management (RCM) and physician practices (independent and healthcare system-based) to tailor Aidéo’s medical coding capabilities to domains through expanded product functionality.• Initiated and executed Large Language Model (LLM) research effort demonstrating how they can complement and extend Aidéo’s current models’ state-of-the-art abilities and further expand the capabilities the platform delivers, leveraging open source foundation models with and without fine tuning, prompt engineering, and RAG.

Provide boutique consulting on applied ML and strategic challenges, with healthcare applications including:• Demonstrated to a commercial health product company that noise in their sensor's readings made the algorithm they employed a poor match to quickly predict a final value. Identified a better-suited alternative algorithm.• Analyzed current applications of home automation technology and developed new service offerings for a telecommunications client to safely and energy efficiently apply this functionality to remote medical monitoring.• Applied and developed machine learning techniques to detect and recognize speech based on surface electromyographic (sEMG) measurements.• Analyzed 'forced cough' data and algorithms to assess potential for early identification of COVID infections using Deep Learning models

Boston, Massachusetts, Us

Ran a Data Science Center of Excellence, improving all business functions for this leading Internet of Things (IoT) company, including:• Defined and oversaw the delivery of a roadmap of metrics, reports, dashboards, and data products to grow high-margin revenue products while supporting internal operations.• Created dashboards translating operational data on 300K+ deployed sensors and tailored for effective rollouts, monitoring, remediation, and executive oversight tasks.• Developed and ran predictive models for equipment failure enabling proactive replacement.• Provided sales engineering capabilities to rapidly prototype data products meeting specific customer needs and securing landmark deals.• Used modeling to demonstrate how to transform a net loss customer to profitability through a new installation strategy; provided monitoring tools to oversee and validate.• Enabled operations team to scale up footprint under management with no increase in staff by implementing a data warehouse that standardized access and metrics across four formerly independent IoT platforms.• Established internal and external (Harvard, Notre Dame) research and development efforts applying Machine Learning (ML) techniques to key challenges, including Deep learning algorithms for classifying equipment events and predicting battery lifetime remaining.• Developed and delivered a stream of blog posts, webinars, and conference presentations to demonstrate product abilities and vision in coordination with Marketing and Sales functions.

Launched and grew the analytics team for this IoT startup through acquisition, transitioning them from an e-commerce sensor hardware company to an Industrial Internet of Things (IIoT) leader and acquisition target. Key contributions included:• Recruiting, hiring, developing, and leading a heterogeneous team to work across internal and external deliverables.• Implementing multiple data warehouses to reduce the complexity and timeline for applying common types of queries needed by the analytics team and other business units to efficiently execute their tasks.• Led data reviews with customers to demonstrate untapped value in their data, building customer-specific data products and data pipelines to meet their individual business needs, and transitioning multiple instances into standard revenue-generating product features.• Enabled effective identification, mitigation, and prevention of operational outages through automated analytics.• Built a range of data products, including an automated classification system to detect problematic refrigeration units and suggest corrective actions or flag specific units as unsuitable for maintaining viable pharmaceutical products.• Demonstrated to regulatory agencies the benefits of virtual product monitoring to model products in dynamic environments and more accurately determine when environmental variations threaten product viability and safety.• Reduced false alarms by implementing normalcy analysis and anomaly detection capabilities in python as the basis for characterizing and predicting defrost events. Used Apache Spark to scale the applications for enterprise customers.• Effectively eliminated field incidents of early battery failures through the introduction of a manufacturing cold soak test, identified and designed through data mining.

Developed health monitoring and coaching capabilities to positively impact user health by providing continuous, real-time vital signs monitoring from a wrist-based biosensor. • Built and led the analytics team as needs evolved from algorithm development to validation to combining biometric data with contextual data sources to generate insights, which increased the users’ awareness and understanding of their body's condition, and the impact of various activities on their health.• Developed algorithms for estimating physiological vital signs from wearable sensors.• Organically developed (i.e. hacked together with Matlab, R, scripting, chron jobs, ...) a back-end system that automated processing of incoming data from prototype devices, applied the latest versions of algorithms under development to the data, compared results where benchmark or reference data was available, and generated accuracy metrics and dashboards.• Led initiative to migrate to a cloud-based back-end, scaling multiple orders of magnitude in data volume as well as supporting the various analytics performed on the data. The design incorporated considerations of the business needs across the algorithm development, device production, clinical research, customer service, and executive teams for monitoring and interpreting data from and about the fielded devices. • Served as a translator from the User Experience team’s vision of product features and capabilities into Research and Development programs through testing and validation.• Designed an Expert System (implemented in Drools) to incorporate clinical Subject-Matter Expert (SME) standards of care and clinical guidelines into a behavior change framework aimed at getting users to adopt healthy habits.• Represented Quanttus in discussions exploring how to work with strategic partners, identifying and defining game-changing opportunities to harness consumer wearable device data, aimed at driving clinically meaningful, measurable health outcomes.

Cambridge, Ma, Us

Worked closely with the bioinformatics team across a range of projects to develop and apply statistical approaches for examining gene transcription data to characterize gene regulation, stratify patients for disease risk, predict outcomes for drug therapies, and identify molecular targets for new therapies and diagnostic tests, including:• Provided analytics services to a global drug vendor, transforming qualitative regulatory gene control knowledge from the public domain into graphical models of gene control pathways for use in determining pre-symptomatic cancer signatures. • Validated models with public domain transcription array data after using R packages to ingest, clean, and adjust data for systematic inter-test and inter-experimental differences.• Integrated software tools from academic partners to apply the latest research methodologies for finding subtle indicators of regulatory relationships in large-scale data sets. • Developed capabilities briefing materials for business development efforts after performing systematic reviews of customer and internal projects. Used materials to re-engage a major client and explore new areas of collaboration after a change in their leadership.

Used a wide range of machine learning techniques to create a virtual coach that helped people adopt behaviors to improve their health and wellness. The “coach” learned from its past interactions to optimize its impact on new users. Key contributions included:• Combining team leadership and hands-on development to drive the creation of the virtual coach’s intelligence, incorporating data analytics tools, and metrics to evaluate individual and population-level impact.• Assumed QA responsibilities for a disastrous contractor deliverable on customer reporting, then redirected project efforts to produce a clean data warehouse and Tableau for the final representation to drive successful delivery.• Contributed to numerous proposals, RFP responses, and business development discussions to represent how the current platform met needs or could be extended to deliver desired capabilities.• Implemented the virtual coach, which delivered user interactions designed by Behavioral Intervention experts, nudging individuals to adopt small behavior changes that cumulatively provide strong positive health and wellness benefits; oversaw 24/7 operations for the Healthrageous population.• Led data analytics efforts to evaluate how specific coach interactions and product features predict engagement levels and health outcomes based on user characteristics.• Developed the hypothesis testing infrastructure to assign groups of users with different coach experiences, allowing optimization of coaching to user types.• Contributed to the User Experience design and planned steady progression of capabilities that enhanced the virtual coach capabilities in step with increasing platform and data analytics capabilities.• Developed the "Durable Blood Pressure" metric to reflect the underlying blood pressure of individuals (after reducing measurement and diurnal noise elements), which served as an easier basis for doctors and clinicians to define simple rules to drive coaching interactions.

Us

Worked with Ab Initio's amazingly scalable and fast software to:• Develop a solution for a telecom carrier that compares monthly Call Detail Records (CDRs) against the CDRs of carriers terminating their calls, helping to determine which bills to refute and providing all the detailed CDR comparison information needed to validate the dispute. This was a high-profile, high-paced implementation that provided a fundamentally new capability for the client to validate monthly charges, putting them on solid footing to dispute billing issues with detailed, call-by-call discrepancy information.• Prototyped a testing tool to evaluate the implementation of a new database against the analyst-defined requirements. This system checked that the right structure existed (e.g. the right fields in the right tables using the right data types) and validated the implemented mapping from data sources (files or other tables) to the new target, generating a report that listed discrepancies between the design definition and the implementation.• Developed an in-house tool for characterizing data held in complex XML documents. This tool parsed XML files into multiple constituent segments, storing these segments in individual files, and then profiling the data contained across the files to provide analysts with the ability to quickly understand the constituent data.

Issy-Les-Moulineaux, Île-De-France, Fr

Headed Orange Labs' efforts in “eHealth” – the application of IT and networking to deliver services in professional and personal health & wellness areas (this group was planned to triple in size in 2010, but the Boston Lab was closed as part of a global cost-cutting effort):• Led an internal consulting group for Orange/France Telecom Group, supporting business units by providing expert advice, designing apps and services, and developing prototypes (to provide tangible demos and assess technical risks). Also supported Orange Business Services in marketing development and operations services to medically oriented companies.• For example, the team developed a personal health and wellness service prototype for the iPhone (and later ported it to Android) that allowed individuals to identify and track their health goals. We used third-party sensors to collect data (e.g. heart rate, weight), journaling to record diet, exercise, and 'good habits,' leveraging social media for encouragement and challenges; this was also integrated with Google Health PHRs to upload the information created in the app and make it available to doctors. A web app allowed users to configure and review their data, and also to share data with their caregivers, whether doctors, family, friends, or hired “coaches.”• Developed the concept for a voice microblogging service to assist in disseminating health information to populations with high illiteracy rates, which was appropriate for NGOs addressing crises or trying to increase general health awareness.• Managed eight simultaneous projects, creating a track record for successful deliveries on schedule and within budget, helping to secure the organization’s reputation as the internal go-to resource in the eHealth area.• Documented the full portable mobile health and wellness prototype, identifying technical risks and key tasks to use it as the basis for a deployable service (originally scheduled to deploy at the end of 2010).

Managed a growing group (30+) of engineers, achieving sales and funding goals while developing core competencies to gain actionable information from large volumes of varied data. This included overseeing a portfolio of projects at all stages from pre-solicitation customer engagement to delivery and follow-on opportunity development. The team developed machine learning and pattern recognition algorithms to quickly make sense of information, overcoming challenges such as noise/clutter that can hide relevant information and varying degrees of certainty in the information (including the possibility of it being incorrect). In addition to group oversight, assumed key roles in specific projects, gathering requirements, developing algorithms, designing UIs, system engineering, designing software, planning releases, and test planning. Some specific accomplishments include:• Led successful proposals for multiple projects (valued at up to $3M) and contributed to others with a total value of over $20M, creating white papers, proposals, and business development discussions.• Delivered ground demonstration of a Multi-Source Data Fusion system as Technical and Project Lead from project initiation ($8M scope effort).• Contributed to accreditation as Capability Maturity Model (CMM) Level 3 by helping define software process, leading one of the pilot projects implementing the process, and successfully undergoing an audit.• Successfully integrated BAE technologies in the areas of tracking, situational awareness, and intelligence analysis in a multi-contractor system, providing counter-IED capability for research evaluations and flight trials. • Exceeded success criteria of DARPA contract for speech recognition based on muscle activity by determining an effective combination of parameterization and machine learning approach, securing three-year follow-on funding to overcome limitations of acoustic-based speech recognition systems in loud environments (factories, aircraft).

Led a team of Ph.D.-level researchers and junior- to senior-level software engineers. Responsibilities included: • Oversight on projects involving applications and extensions of AI techniques ranging from Bayes Nets to Expert Systems.• Oversaw the individual development of the section members as well as the creation of team camaraderie. • Supported Marketing efforts (e.g. proposal writing, collaboration evaluation) to ensure new and continued funding of projects to support the section and allow for steady growth.• Interacted with other sections, groups, and divisions, providing an internal resource of knowledge related to probabilistic reasoning systems.• Developed staff planning to match project needs within and beyond section efforts with the skill sets and career objectives of section personnel in a matrixed organization.• Created schedules and budgets for projects ranging from $70K to $3.3M and subsequently monitored the projects by schedule and budget.• Designed and implemented a system that automatically translates a set of awareness goals into a set of input collection needs and automatically interprets collected data to assess the situation and the accomplishment of goals.• Led development and implementation of techniques for Data Validation, identifying problematic data based on consistency with contextual information and historical accuracy of sources.• Executed development, implementation, and application of techniques for Value of Information calculation, determining what collectible data had the greatest potential to resolve ambiguities and improve mission success.• Led integration of an Expert System with a normalcy learning and anomaly detection system for monitoring maritime shipping information to detect safety and security issues.

Performed numerous roles across Data Fusion and applied AI projects, including:• Design and implementation of a Bayesian Network for use in detecting vehicles whose movements within a battlespace are consistent with the movements of vehicles engaged in Theatre Ballistic Missile (TBM) activities. • Extension of an internal Bayesian Network software tool (C++) to allow for Multi-topological Dynamic Bayesian Networks (DBNs) to improve the TBM vehicle model described above, along with significant debugging of the tool's implementation for an approximate inferencing algorithm and suggestions for changes to its implementation to improve speed and robustness.• Development of a convoy detection algorithm and oversight of its implementation in C++.• User Interface design to make the TBM and convoy detection algorithms accessible to forward-deployed military personnel.• Development of an approach to transform Bayes Nets used for classification in the missile defense domain into a predictive form, enabling sensor resource management with extended lookahead horizons.

Initiated and led Emerging Technologies Group within Iperia, which included:• Prototyping/validating integrations of new technologies and/or products with Iperia's telephony application platform.• Serving as a technical resource for Business Development, both to introduce technical understanding of the product into business development discussions and to enable effective communication of industry trends back to engineering.• Developing demo systems for use at trade shows and as sales tools.• Monitoring standards relevant to future work.The process I introduced for projects within this group, including short-term assignment of engineers from outside the group to work with core engineers, was designed to spread knowledge and experience of new technologies within the Emerging Technologies group and the Iperia organization. Accomplishments within this group include:• Leading the development of a voice-driven Auto Attendant application, using a GUI tool to enable an administrator to configure hours of operation and edit the employee data set, which produced XML descriptions of the data itself, then using XSLT to transform the data into VoiceXML describing the AutoAttendant call flow.• Successfully produced a demo of the integration with a third-party IP Conferencing product under tight time constraints for an industry trade show.• Designed and oversaw the development of integrated GUI control for conferencing functionality, customizing the look and feel to match that of Iperia's core offering.

Involved in all aspects of design and development for speech-enabled telephony applications, with duties including:• Application Software Design and Development (C++)• Specification Writing• User Interface Design• Customer Relationship Management• Demo Development and Support• Recognition Grammar Development with Natural Language Shortcuts

Implemented Jordan's Hierarchical Mixture of Experts Architecture in C programming language for phoneme probability estimation. Presented work with a spotlight poster at the 1994 Advances in Neural Information Processing Systems (NIPS) conference.

Led an effort to determine if high magnetic fields and electromagnetic pulses involved in Magnetic Resonance Imaging (MRI) would cause hazards for subjects with Cochlear Implants.

Developed an artificial intelligence system for use in real-time monitoring and control of fermentation. Conducted a number of fermentations to create a knowledge base and trained a neural network to classify fermentation states for use by an expert system. Tested the system by running fermentations and introducing various faults. Constructed desktop-scale (<2.0 Liter) fermentation system.

Ph.D, Speech And Hearing Sciences

Bachelor Of Science, Electrical Engineering

Bachelor Of Science, Managerial Science