Sharath
Kulkarni
Building applied AI systems in Healthcare
I work inside a healthcare organisation & MSO where care delivery, operations and revenue all meet. Most days are spent aligning clinicians, front desk teams, billers and engineers around the same problems.
Over the past few years that has meant shaping an internal platform that follows a patient from first contact to final payment. Referrals, intake, eligibility, scheduling, program workflows, claim status, denials and leadership analytics now share one set of rails, with early voice and text AI layered on top.
This site is a quick tour of that work and a way to show how the same mix of provider side experience, MSO operations and product leadership can help other teams who are trying to build practical AI in healthcare.
Path and roles so far
AI and systems lead for a specialty network
Works closely with physicians, nurses and clinic staff across multiple sites in a specialty network.
- Shapes how referrals, programs and visits actually move through the day, from first contact to follow up.
- Designs analytics that give leaders a clean view of access, throughput and value based care performance.
- Experiments with AI where it helps most, such as summarising information, highlighting risk and supporting conversation-based tools, while keeping clinical teams in control.
TAKEAWAY
"Comfortable translating messy front line reality into roadmaps and systems that both clinicians and product teams trust."
Understand the ground reality first, then design systems and stories that respect it.
What is being built right now
Building a centralized AI-powered layer that sits on top of existing tools so intake, coverage, visits, billing and learning all feel like one connected system instead of silos.
First contact
Referrals, calls and digital requests land in one place instead of being scattered across fax queues, inboxes and sticky notes. Repetitive intake questions are handled by simple automations so staff can focus on people, not paperwork.
Coverage and readiness
Eligibility, benefits and basic checks are pulled earlier in the journey. The system flags gaps in coverage, missing authorizations and high-risk cases before the visit, so there are fewer surprises for patients and fewer write-offs for the practice.
Visits and programs
Visits, procedures and longer-running programs show up as a clean timeline across locations. Notes, orders and follow ups are supported by lightweight voice and text tools, so what actually happened in the room is what shows up in the record.
Turning care into data
Charges, codes and program enrollment are captured once, in a structure that flows cleanly into billing, quality reporting and value-based contracts. Rules live in one place, so teams do not have to remember every edge case for every plan.
Money in motion
Claims go out clean, and status comes back into a single view instead of a dozen portals. Worklists are generated automatically, routing issues and next steps to the right people instead of another shared spreadsheet.
Learning from the loop
Denials, underpayments and performance trends roll up into simple views that point to what needs to change in the system – front-end checks, documentation, coding or follow up – instead of just adding more tasks to the queue.
System Snapshots
Giving payer & vendor contracts a home; the team actually uses
"Rates, terms and renewal dates lived in PDFs, email threads and paper folders. Underpayments were spotted one claim at a time."
Introduced Nexus, an internal contract and rate layer that stores fee schedules, terms and key dates in a structured way. Connected Nexus to ClaimCatalyst so every claim carries its expected allowed amount and basic rules for that plan.
Underpayments and bad terms are no longer a guessing game. Difficult or unprofitable plans rise to the surface, and leaders can see which contracts need attention without starting a months long forensic project.
Making denials less mysterious
"Denials and follow up work lived in portals, spreadsheets and personal notes. Patterns were hard to see."
Brought claims, reasons and actions into one view with simple categories that operators actually use.
Leaders can now ask why a plan is difficult and get an answer that is backed by data, not just frustration.
Trying conversational AI where it helps
"Teams and patients spent too much time inside rigid phone trees and scripts that did not respect context."
Started layering voice and text AI on top of existing workflows for routine questions and status checks, with clear boundaries.
Early signs show less friction on simple tasks and more time for staff to handle the hard ones.
Education & Training
MS in Computer Science
[Sofia University – California]
Focus on AI, data and software engineering while working full time in health care operations.
Master’s degree
[California State University, Fresno]
Advanced work in business, analytics, and operations.
Bachelor’s degree
[Bangalore University]
Built core skills in marketing, finance, and management.
Additional certifications
Professional Development
Independent certifications in product management, marketing, AI and data visualization.