Skip to content
View nandanito's full-sized avatar

Block or report nandanito

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
nandanito/README.md
Nandan Joshi — physics → machine learning → quantum

Physicist by training, now working where machine intelligence, quantum computing and cryptography for digital assets meet. What ties them together for me is how information becomes structure, signal, and eventually knowledge.

Now

QMI Lab · the independent lab I run
Research on intelligence and learning across classical and quantum computation. Current work: hybrid quantum-classical Transformer fine-tuning, and cross-lingual transfer through romanization.

AstroLLM · building an LLM for astronomy
Retrieval-grounded, QLoRA-tuned on Qwen3, wired to NASA ADS, SIMBAD and the Exoplanet Archive. Evaluated on grounding, tool routing, abstention and teaching quality.

Agent infrastructure · open source from QMI Lab
Lodestar, a trust layer for AI agents: what an agent believed, why it acted, and whether it was right. Asterism, a local-first runtime for many distinct agents, each with its own memory, secrets, skills and autonomy.

Confidential Digital Assets · privacy on public ledgers
What confidential-token systems actually hide, established by running them rather than reading their docs, with every claim traced to evidence. Two studies so far: XLS-96 on the XRP Ledger (write-up) and Zeto on Paladin (EVM).

PQC · post-quantum cryptography
Implementing post-quantum schemes from their NIST standards, and explaining them along the way. First project: slh-dsa-zig, SLH-DSA (FIPS 205) in pure Zig, with a guide to how the scheme works.

CrossMono · open e-reader firmware
Reading-first firmware for the M5Stack PaperMono e-paper devices (ESP32-S3), built on CrossPoint Reader and the FreeInk SDK, with changes contributed upstream to both. Installs from the browser at crossmono.github.io.

Beyond the lab, I work across the wider quantum computing stack: algorithms and circuits, hybrid quantum-classical methods, simulators and hardware back-ends, and the developer tooling that connects them.

Also learning in public: array-thinking-to-q, from J to q/kdb+ with every example executed.

Writing

More at nandan.me/writing.

Background

  • Diplom Physik, Göttingen (2007). DFG-funded doctoral research in stellar astrophysics at the Institut für Astrophysik Göttingen: observing at ESO facilities in Chile, custom IDL pipelines for spectroscopic analysis.
  • Max Planck Institute for Dynamics and Self-Organization. Q-learning agents controlling simulated artificial muscles, with the biomechanics modeled through Lagrangian mechanics. Reinforcement learning from before the deep-RL era.
  • Publication: Reiners et al., A Catalog of Rotation and Activity in Early-M Stars. The Astronomical Journal 143, 93 (2012). ADS

Toolbox

Python PyTorch Hugging Face vLLM Unsloth PennyLane Qiskit Zig C++ TypeScript q/kdb+ PostgreSQL Astro

Elsewhere

nandan.me · qmilab.com · astrollm.org · LinkedIn · X

Pinned Loading

  1. crossmono crossmono Public

    Reading-first community firmware for the M5Stack PaperMono and PaperMono-Lite, built on CrossPoint Reader and the FreeInk SDK. Not affiliated with M5Stack.

    C++ 3

  2. qmilab/lodestar qmilab/lodestar Public

    The trust layer for AI agents.

    TypeScript 5

  3. astrollm/astrollm astrollm/astrollm Public

    A domain-specialized Large Language Model for Astronomy & Astrophysics

    Python 3

  4. xrpl-confidential-positions xrpl-confidential-positions Public

    What XLS-96 Confidential MPTs actually hide on the XRP Ledger, established by running a reference application. Devnet only.

    TypeScript 1

  5. array-thinking-to-q array-thinking-to-q Public

    Array thinking, all the way to q — a J laboratory then q/kdb+. Learning in public.

    Python 1

  6. slh-dsa-zig slh-dsa-zig Public

    Pure-Zig implementation of SLH-DSA per FIPS 205. Phase 1 of pq-zig. EXPERIMENTAL.

    Zig 1