Independent researcher · AI safety · evals · quantum-informed ML
Kishore
Sakthivel
I run small, falsifiable experiments on one question: how little do you need to observe to know enough? The cloud behind this page is a 2,400-amplitude state, and only the bright points are measured.
Read the hypothesisGet in touchby my AI systems
Spatial Parse
non-AI engine
Current research
Measure a little.
Learn a lot.
Can compact measurements of high-dimensional quantum states provide useful latent representations for an existing AI architecture, improving generation, search, or verification without requiring explicit reconstruction of the full quantum state?
Protocol · no results claimed yet
Kill conditions
come first.
State families, measurement schemes, baselines: raw state, random projection, PCA.
Search accuracy vs number of measurements under shot noise.
Adapter on the latent for a frozen generator, with a parameter-matched ablation.
Drift detection vs fidelity estimation at equal shot budget.
Public write-up with code, seeds and negative results.
Why this counts as AI safety
Claims that
can fail.
Shipping is my research taste · tap to open
Systems
that run.
Resume Engineer
Spatial Parse
TermsGuard · ATS Optimizer
Path
Engineer first,
then researcher.
Independent Researcher, AI systems and quantum-informed ML.
Freelance writer: CVs, ATS resumes, cover letters.
Operations Associate, Flipkart Seller Hub, Bengaluru.
Freelance web developer: Python, APIs.
B.E. Computer Science, Arulmurugan College of Engineering. CGPA 7.99.
Open to fellowships, collaboration and critique
Try to
break it.
The best reviewer is the one who attacks the hypothesis.
Thanjavur, India · open to relocation