Rishabh Raj / Engineer reaching for research
I turn research ideas into instruments that survive contact with data, hardware, and the real world.
Scroll to observe
Three systems. One habit: make the idea executable.
A paper made executable: condition monitoring from signal to silicon.
Nine-level Daubechies-44 feature extraction, ANN inference, fixed-point export, board-independent RTL, and MagNet motion amplification—held together by bit-accurate Python/Verilog verification.
9 → 32 → 64 → 128 → 64 → 32 → 1Inspect the research systemA research pipeline for turning market structure into reviewable decisions.
At NorthRock Capital, I built resilient ingestion across 250+ instruments, a local-LLM research workflow, and a constrained, monthly-rebalanced QMI portfolio model for a $1M initial allocation.
$450M+ AUM · 250+ instrumentsRead the documented workAgricultural intelligence engineered for the messy edges of production.
A serverless React/Node platform with bounded crop inference, searchable public schemes, protected administration, Firebase routing, accessible states, and a 96-tree model embedded directly in the API runtime.
96-tree inference · bounded APIsOpen the systemA signal becomes useful when its assumptions become visible.
Research judgment with an engineer’s intolerance for hand-waving.
Paper → instrument
Reproduce the claim, expose the assumptions, design the controls, and make the result falsifiable.
Signal → decision
Build research pipelines where provenance, constraints, regime sensitivity, and reviewability are first-class.
Model → reality
Carry ideas across APIs, cloud data, hardware boundaries, reliability checks, and human operations.
“A model is not finished when it predicts. It is finished when its assumptions, failure modes, and route to reality are visible.”
Proof that the range is real, not rhetorical.
JavaScriptSecurity-hardened MERN commerce with server-authoritative pricing and idempotent Stripe fulfillment.
02TypeScriptA deterministic browser experiment where spectral agents move through superposition, entanglement, and decoherence.
03PythonVerified, from-scratch machine-learning implementations with an emphasis on mathematical primitives.
04C++A working archive of algorithmic problem solving in modern C++.
From physical systems to financial ones—and toward research.
I am drawn to problems where the signal is faint, the system is consequential, and the answer has to work outside a notebook.
My route is deliberately non-linear: electrical engineering, competitive programming, production operations, quantitative development, and financial engineering. The destination is AI research or quantitative research with real technical depth.
Download résuméStudied the boundary between computation and physical systems; graduated with a 7.79/10 CGPA.
Built investment-research infrastructure, portfolio intelligence, and constrained quantitative models for a $450M+ AUM team.
Deepening the statistical, econometric, and mathematical foundation behind market research.
Own high-availability LPG operations at 300 MT/day, automate recurring workflows, and lead multidisciplinary reliability work.
Five independent records. Each opens its source.
For research, systems, and difficult questions worth staying up for.
Let’s find the signal.
rraj09102002@gmail.com