Journey · 2024—2028
An idea
before evidence.
Before execution came formulation: patient reading, a precise question, and the willingness to build evidence around it.
From isolated experiments to research systems
What began as an attempt to see where a PINN was failing became a study of how an adaptive system should intervene—and when it should undo itself.
The work grew through failed paths as much as successful ones. A formulation was reverted when it changed the problem. A sensitivity sweep was removed when its configuration had not propagated. These were not interruptions to the story; they made the story scientifically useful.
Independent experiments
Moved from Burgers-equation PINNs into full two-dimensional Navier–Stokes modelling and field-level diagnostics.
Follow this thread ↗An idea before evidence
Formulated a question around hidden variable–region failures before having a benchmark result to defend it.
Follow this thread ↗Research fellowship
Selected for the Science Academies’ SRFP and began turning the question into a reproducible programme.
Follow this thread ↗A guarded system
Built VARA-PINN through reverts, multi-seed studies, ablations, budget guardrails, and a full-state rollback mechanism.
Follow this thread ↗A second systems direction
Developed and submitted SafeGraph-CAV, joining learned relational trust with cooperative control and a separate physical shield.
Follow this thread ↗Evidence into manuscripts
Consolidated research artefacts, statistical summaries, integrity checks, and the VARA-PINN journal submission.
Follow this thread ↗Expected completion
B.Tech CSE completion; a future milestone, not a completed achievement.
Follow this thread ↗
“A fellowship created the space to turn a question into a reproducible research programme.”