- 🛡️ Currently building AI security at Unbound Security (YC S24) — secrets/DLP detection, BYOC deploys on Amazon EKS, and production observability.
- 🤖 I build and study ML systems — training language models from scratch, benchmarking LLM adversarial robustness, and the backend & cloud infrastructure that ships them.
- 🎓 Integrated M.Sc. Data Science at PSG College of Technology (expected 2027).
- 💬 Open to AI/ML and backend / SRE roles.
| Project | What it is |
|---|---|
| redharness | Standardized, reproducible benchmark for LLM adversarial robustness — jailbreak, prompt-injection & data-leakage under one methodology. · PyPI |
| nanocoder-math | A decoder-only GPT trained from scratch for code & math under single-device compute, with controlled ablations. |
| RAMPART | Regime-aware multi-agent RL portfolio allocation — built evaluation-first with purged walk-forward backtests. |
| launchpad | Self-hosted PaaS that deploys apps from source into your own AWS account via Terraform. |
| syncset-db | Selective PostgreSQL replication with schema-drift detection & crash-safe sync. · PyPI |
| codegraph | Turn any GitHub repo into an interactive code knowledge graph with semantic search. |