AI security · governance · systems thinking — building at the intersection of research and practice.
Project Aegis / TRiSM_OS — an applied governance fabric for AI systems in the wild: policy-as-code, context-aware guardrails, and post-deployment monitoring. The work threads research into behavioral and systems constraints into shippable controls.
OSS contribution pipeline — upstream fixes where they matter: microsoft/presidio → PII anonymization correctness, BerriAI/litellm → guardrail validation for TRiSM context. Only merged PRs are shown above.
The contribution graph above is a neural metaphor applied to contribution data — a network of nodes that fire and settle, not a dashboard.
- Graphite — a dormant day (no contribution that day)
- Cyan-white — the firing flash when a day activates
- Plum — the resting/strengthened state: active days settle into a glow brighter than dormant, weighted by contribution intensity and recency. This residual trace — the neuroplasticity effect — is what makes it read as a network with memory rather than a recolored graph.
One chronological sweep moves left to right; dense weeks wash the grid with a soft violet; the whole network resolves and breathes once before resetting. It's a metaphor for how systems (and people) accumulate and hold state over time — no streak counters, badge walls, or skill grids.
Interests: AI security, governance, systems thinking, building, research and behavioral science of how systems (and people) behave under constraints.
- Portfolio:
https://arjunpakhan.example.com(placeholder — wire to future Obsidian Brain site) - LinkedIn:
https://www.linkedin.com/in/arjunpakhan
Static SMIL-animated SVG via generate.py (no JS/WebGL), regenerated daily by Actions to the output branch.