Projects
Work grounded in products, evidence, and experimentation.
Selected work across AI-powered product exploration, data systems, verification, and autonomous-agent experimentation.
Co-Founder · Adaptive AI Data Intelligence
NeuroNova
NeuroNova is an Adaptive AI Data Intelligence startup building toward a persistent AI data analyst that progressively understands each company's unique data language, context, and analytical workflows. As Co-Founder, I work across product and AI architecture while helping refine the MVP through real-world validation.
Flagship project
RoadBack · a founder story
Founder story · 2025 – 2026
RoadBack
Built from my own experience recovering from a serious accident, RoadBack is an AI-powered recovery platform designed to give orthopaedic patients more structured and personalized guidance throughout recovery.
View on GitHub (opens in a new tab)- Origin
- Built from my own experience recovering from a serious accident.
- Problem
- More structured and personalized guidance throughout orthopaedic recovery.
- Product
- An AI-powered recovery platform for orthopaedic patients.
- Technical direction
- An AI-powered product direction; no clinical claims or outcomes are presented here.
- Early validation
- Conducted limited conversations with physiotherapy clinics around the problem and patient adoption.
- Current status
- Founder-led exploration and early validation work during 2025 – 2026.
In development
Adaptive Compute Lab
ML research · In development
Adaptive Compute Lab
An early-stage research project exploring whether validation-calibrated adaptive exits can improve the accuracy-latency trade-off of looped Transformers on controlled tasks. I am currently building the foundations and baseline experiments.
- The six-month plan starts with core ML foundations, a verified task generator, and a trusted conventional baseline before comparing fixed and adaptive computation.
More work
A growing collection of systems and product experiments.
Buildathon agent
Change Impact & Verification Copilot
A buildathon agent that combines code-impact evidence with runtime traffic to determine what should be verified after code changes.
- Generates a Change Impact Report with Confirmed, Heuristic, and Needs Verification categories.
Strategy evaluation platform
Quantive
An autonomous quantitative strategy-evaluation platform and data layer for market-data ingestion, storage, and analysis.
- Built using PostgreSQL, TimescaleDB, FastAPI, SQLAlchemy, Alembic, Docker, and asynchronous workflows.
Autonomous agent experimentation
Trading Environment
A reinforcement-learning environment for autonomous trading agents and strategy experimentation.
Personal finance tooling
Personal Finance Dashboard
A React + Firebase dashboard for finance and investment tracking and visualization.