RESEARCH / MARKETS
Market structure
Indian equity and options data, a backtesting harness, and a record of ideas that survived—or didn’t.
Open the notebook
A small lab for
big questions.
Systems, research, and things you can play with.
Selected work
SELECTED WORK
PRACTICE / OPERATIONS
Business systems
Understanding real processes, structuring the data, and building or configuring the tools that make the work usable.
Explore the practice
01India from orbitSatellite data, joined on a five-kilometre grid. Open the research or explore the map.Geospatial research
02Research OSA research system with baselines, external checks and a referee for its own findings.Research systems
03A city of six shapesA street built in Unreal Engine, a finished film, and the notebook of its making.Film + real-time rendering
04The market research hubThe original captured workspace: scanners, breadth, research and a labelled paper book.Historical workspace
RECENT EXPERIMENTS / OCTOBER 2026
Less watching.
More playing.
Working browser experiments, with their rough edges left visible.
Open the playground
INTERACTIVE / AUDIO + VISUALS
Resonance
A mechanical reactor that charges, opens and takes you inside. Play the original music set, load your own song, or direct the lights and camera.
Layered artwork + a 3D interior. Song cues are estimates. Audio stays on your device.
Launch the instrument
COMPUTER VISION / PROTOTYPE
Your hands move it.
Two hands control size and rotation through your camera. Direct geometry, a visible response, and local processing.
Try hand control
COMPUTER VISION / RESEARCH TRIAL
See the movement.
Inspect player detections and trails in a badminton clip. Sampled movement tracking, with misses and broken IDs visible.
Run the movement trialSMALL SYSTEMS / WORK SAMPLES
Useful, before elaborate.
Illustrative work, with fictional inputs. These show the method; they aren’t client outcome claims.
Enquiries, ownership & the next stepAn interactive CRM demonstration with fictional solar enquiries. Assignment, follow-ups and an inspectable history.Try the CRMEngineering budget estimatorVersioned rates, work breakdown, expenses and visible reconciliation in a spreadsheet.Explore the sampleContact cleanup with a paper trailBefore and after data, conservative duplicate rules, review flags and row-level lineage.Inspect the process
FIELD NOTES
What these experiments taught me.
A few distinctions that matter more than a polished demo.
Reactive visuals don’t need an AI model.
Resonance uses Web Audio frequency energy and a scored sequence. Bass, mids and hits feed explicit visual controls. Local-song build and drop cues are energy-based estimates, not an understanding of the track.
Detection isn’t coaching.
A person detector can produce boxes and image-space trails. Tactical analysis, court distance and technique correction need calibration, stronger tracking and validation. The badminton trial deliberately stops at observations.
Keep the experiment that challenges the story.
The original money-exchange simulation is still here: identical agents, random exchanges, conserved total. Watch a distribution emerge, then compare it with the mathematical prediction.
Run the exchange modelABOUT / CONTACT
Bring a good question.
I’m Nandit—a computer-science engineer with a systems and analytics mindset. I work across business operations, research and practical experiments.
The business practice lives at Lantern. This is where the wider work lives.
vaishnandit@gmail.com