Skip to content
View adarshsm's full-sized avatar

Block or report adarshsm

Report abuse

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

Report abuse
adarshsm/README.md
Typing SVG


Five years shipping AI from research to production.

Currently building AURA — a 30B multimodal foundation model generating 2K video + synchronized stereo audio in a single forward pass, with sampling cut from 50 steps to ≤8 via Rectified Flow + step distillation.

"A model that only works at fifty sampling steps is a research result, not a product."

The through-line across all of it: inference budgets — getting capable models to run fast, cheap, and reliably outside the lab.



🔭 What I'm Building

🎬 AURA

30B multimodal foundation model. 2K video + stereo audio in one forward pass. 50-step sampling distilled to ≤8 via Rectified Flow + Latent Consistency Models.

PyTorch DiT Rectified Flow Distributed Training

📜 Sigil

Mobile app converting spoken agreements into binding micro-contracts. The LLM runs entirely on-device — zero cloud round-trips, complete privacy.

On-device SLMs Speech Processing Privacy-first

SemantiCache

Semantic LLM caching. 61% cache hit rate on real chatbot data. Cut API costs 40–70% with one Python decorator. Zero external dependencies.

Python Sentence Transformers SQLite MIT


🌍 Open Source

Correctness fixes over cosmetic changes — every fix ships with a failing-then-passing test, and each one is reproduced and root-caused before a line is written.

8 merged upstream · 11 more in review

Repository Stars Contribution Status
k6 31k+ ⭐ Windows path separators corrupted browser file-persister URLs ✅ Merged
recharts 27k+ ⭐ Rectangle unmounted instantly instead of animating down to zero 🔄 In review
matplotlib 23k+ ⭐ wxPython 4.3 opened plot windows collapsed to ~132×131 px ✅ Merged
DVC 15k+ ⭐ Mixed .dvc/path targets silently dropped from pull · params named size/nfiles skipped the run-cache hash 🔄 2 in review
Gymnasium 12k+ ⭐ Text space sampled values its own contains() rejected ✅ Merged
statsmodels 11k+ ⭐ SARIMAX extend() rejected exog that was merely constant over the new window ✅ Merged
Hydra 10k+ ⭐ instantiate() merged call-site dict arguments instead of replacing them · errors advertised HYDRA_FULL_ERROR to users · redundant macOS CI step · project homepages in the Landscape ✅ 2 merged · 🔄 2 in review
folium 7k+ ⭐ Vendored a glyphicons-only stylesheet instead of all of Bootstrap 🔄 In review
Feast 7k+ ⭐ Redis event-time ordering · pymssql arm64 wheels · registry caching + materialization reporting ✅ 2 merged · 🔄 3 in review
sqlparse 4k+ ⭐ ROLE used as a column name stayed a bare keyword 🔄 In review
dulwich 2k+ ⭐ foo/**/ in .gitignore stopped constraining matches to directories 🔄 In review

🧠 Tech Stack

Generative AI & ML

PyTorch Diffusion Transformers Rectified Flow LCM Fine-tuning RAG FAISS Pinecone

Languages & Frameworks

Python TypeScript C++ React Next.js

Cloud & MLOps

Azure AWS IBM Cloud Power BI


📊 Stats


📌 Featured


📍 Bengaluru, India  ·  Open to collaboration on inference optimization, multimodal generation, and impactful open source

Popular repositories Loading

  1. Fingerprint-Based-IoT Fingerprint-Based-IoT Public

    C++

  2. Smart-Home-IoT Smart-Home-IoT Public

    C++

  3. pH-Sensor pH-Sensor Public

    C++

  4. react react Public

    Forked from react/react

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

    JavaScript

  5. CodeCraftHub CodeCraftHub Public

  6. langflow langflow Public

    Forked from langflow-ai/langflow

    Langflow is a low-code app builder for RAG and multi-agent AI applications. It’s Python-based and agnostic to any model, API, or database.

    Python