Tarush Gupta
Research Intern at a Stealth Post-LLM Startup; Co-founder of VeriCare AI; Founder and Host of the Neural Bridge Podcast
About
I'm Tarush Gupta, a researcher and builder currently working at a stealth post-LLM startup and co-founding VeriCare AI, where we use agentic AI to tackle the 80% error rate in US medical billing. My journey has taken me from product roles at Turing and Ema to pretraining my own language models from first principles. I’m deeply passionate about AI alignment, Stoic philosophy, and ensuring that the next generation are co-authors of AI policy rather than just subjects. Through my podcast, Neural Bridge, I aim to connect industry veterans with the next wave of innovators. I’m always looking to collaborate with frontier labs or chat with researchers who are pushing the boundaries of what 'Software as an Agent' can truly become. Onward!
Networking
What I can offer
- ›Insights on LLM data sourcing and post-training
- ›Medical billing/RCM domain expertise
- ›Platform for industry leaders via Neural Bridge Podcast
- ›Youth perspective on AI policy and ethics
Looking for
- ›Industry vets and researchers for podcast guests
- ›Collaborative opportunities with frontier labs
- ›Mentorship from high-level CEOs and founders
Best fit for
Current Interests
Background
Career
Progressed from sales and product management internships to deep technical roles in LLM post-training at Turing and founding an AI-driven medical billing startup, while serving as a scholar and AI ambassador.
Education
The Harker School: High School Diploma candidate (Class of 2027)
Achievements
- ›Pretrained AureliusGPT SLMs from first principles using NumPy
- ›Developed VeriCare AI agentic solution with 20+ workflows for medical billing
- ›Selected as the inaugural Chen Lin Scholar for AI for Humanity
- ›Developed Path to Plane app with 30+ daily active users
Opinions
- Data is King
- Internalizing first principles by building without high-level libraries is essential
- Medical billing is a systemic failure AI is uniquely positioned to solve
- AI safety and youth protection should be a market advantage, not a constraint