Ramchand Kumaresan
AI Researcher & Program Leader (PMP, PgMP) · Austin, TX
Building and publishing AI, on 16 years of delivering programs with no existing playbook.
I'm an AI researcher and program leader. I've authored six research papers on large-language-model systems — on arXiv, with submissions to NeurIPS 2026 and TMLR — and I ship AI products in production. I do it on a foundation of over 16 years delivering complex SaaS and digital-transformation programs across startups, enterprises, and public-sector clients, with both PMP and PgMP certifications from PMI. The throughline is execution where there is no existing playbook.
My domain expertise spans healthcare, construction technology, public-sector systems, and enterprise SaaS. At Cognizant Technology Solutions, I delivered end-to-end implementations in the healthcare domain, including Affordable Care Act programs for Aetna. At Aspira, I led statewide rollouts for campground reservation systems, hunting and fishing licensing platforms, and kiosk deployments — initiatives that directly impacted millions of end-users. Across every engagement, I've owned the full lifecycle from RFP through launch and operations transfer.
I'm deeply data-driven in how I manage programs. I track delivery KPIs including schedule adherence, risk closure rates, and stakeholder satisfaction alongside engagement outcomes like adoption and ROI — and I use them to make decisions, not just report status.
Select Achievements
Led the statewide launch of the Texas State Parks reservation system across 90 parks and 8,000 campsites. Identified an untested WCAG 2.0 accessibility gap two weeks before launch, drove recertification, resolved 37 failures, and achieved AAA ADA compliance on schedule — delivering a system that served 400+ customers and 13.5 million records at launch.
Managed the Keep Colorado Wild Pass initiative across 64 counties, coordinating three state agencies and designing a phased go-live strategy. The program sold over 1.2 million passes in six months, generating more than $35 million in revenue with 99%+ data sync accuracy.
Coordinated Utah's draw system migration across engineering, product, and state agencies — delivering on time and compressing the customer onboarding cycle from two years to six months.
Resolved a critical payment gateway failure on a $12 million Colorado contract within five days by isolating a caching mesh configuration issue across API nodes, averting a two-month delay and a potential $3 million loss. The triage structure was adopted as standard practice.
Navigated a data migration crisis during the Texas launch — 5,000 future reservations incorrectly flagged as historical at cutover. Designed and executed a manual recovery plan with 20 team members overnight, launching on time the next morning.
Delivered a full production migration from on-premise data centers to AWS in 75 days with under 90 minutes of downtime, achieving 40% query performance improvement, ~30% infrastructure cost savings, and zero compliance findings across PCI and SOC 2 audits.
Applied delivery KPIs on a statewide program to reduce post-launch support tickets by 35% while improving customer satisfaction scores by 22%.
My AI work is hands-on, not advisory. I train and evaluate language models on local hardware (an RTX 5090 and a Mac Studio M4 Max), reverse-engineer on-device inference paths, and design multi-agent orchestration systems. Recent results include a +21.76% cross-lingual model-fusion gain (KALAVAI, heading to NeurIPS 2026) and an 8.5× Apple Neural Engine compilation speedup (Orion) — every result reproducible from committed code. I've also shipped AI products in production: a live personalized-storytelling platform and a clinical-triage assistant that runs a 5GB model fully offline on a phone. This isn't a side interest — it directly informs how I assess AI capability, technical feasibility, and build-vs-buy decisions in a leadership role.
Based in Austin, TX. Open to senior program leadership roles where AI fluency, systems thinking, and execution discipline matter.