Hi, I am Samarth.
I am a Research Engineer and part-time French student. I spend my time learning how things work and designing novel solutions to problems I cannot get out of my head. Right now, I am most curious about world models and agents.
I currently work on optimizing models and Neural ODE research. I also support research and development at Seneca Helix as a founding member, focusing on decentralized systems for AI agents.
Samarth Sharma — Research Engineer
Focus: SciML, Neural ODEs, agents, world models
- Current: Neural ODE optimization, research tooling
- Past: Seneca Helix — decentralized systems for agents
Capabilities
- Model optimization & evaluation
- Neural ODEs & scientific ML
- Agent design & orchestration
Selected Work
- Project A — concise description
- Project B — concise description
Work
AI Researcher @ Vizuara
May 2025 - Present
Optimizing Bayesian models for uncertainty quantification
Founding Software Engineer @ Seneca Helix
March 2025 - Present
Working on decentralized systems for AI agents
Software Developer Intern @ Sialka Cloud Solutions
Sept 2024 – Apr 2025
Built and deployed microservices, migrating to AWS
Software Engineer Fellowship @ Headstarter
Jul 2024 – Aug 2024
Collaborated to deliver 5 applications in 6 weeks
Projects
Neural ODE
Research, 2025
Scientific machine learning pipeline that analyzes storage frequency interactions to interpret microgrid control
Shannon
Project, 2025
The stateful web agents that allows user to harness context window
Fragments
Project, 2025
A microservice, Fragments API, for storing, retrieving, and transforming content fragments using AWS
AgriTrail
Hackathon, 2025
A prototype for tracking and verifying food products using Hyperledger Fabric and ESP32
Blog
Antimatters Labs
2025
A note on Physics × AI.
antimatterslabs.com
Software Portability & Optimization — Hardcoding Machines
2025
Notes on writing software that travels across hardware without performance cliffs
Read
From Systems of Equations to Neural Networks: A Linear Algebra Journey
2024
From linear systems intuition to how gradient based nets actually learn
Read