File size: 8,236 Bytes
206752a 77a43a4 206752a 77a43a4 206752a 77a43a4 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 | ---
title: Rishabh Patil
emoji: 馃攷
colorFrom: indigo
colorTo: blue
sdk: static
pinned: true
short_description: AI engineer and researcher. Models, data, papers.
tags:
- portfolio
- claim-verification
- text-to-sql
---
<p align="center">
<img alt="Rishabh Patil. AI engineer and researcher, co-founder and CEO of Valuren." src="https://huggingface.co/spaces/rishhh/about/resolve/main/assets/banner-auto.svg" width="100%">
</p>
<p align="center">
<a href="https://huggingface.co/rishhh"><img alt="Follow me on Hugging Face" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/follow-me-on-HF-sm.svg"></a>
<a href="https://www.linkedin.com/in/rishabh-ashok-patil/"><img alt="LinkedIn" src="https://img.shields.io/badge/LinkedIn-0A66C2?style=flat-square"></a>
<a href="https://github.com/MrRobotop"><img alt="GitHub" src="https://img.shields.io/badge/GitHub-MrRobotop-24292f?style=flat-square&logo=github&logoColor=white"></a>
<a href="https://orcid.org/0009-0007-0868-9673"><img alt="ORCID" src="https://img.shields.io/badge/ORCID-A6CE39?style=flat-square&logo=orcid&logoColor=white"></a>
<a href="https://www.rishabhpatil.com"><img alt="Website" src="https://img.shields.io/badge/rishabhpatil.com-24292f?style=flat-square"></a>
</p>
AI engineer and researcher in London. Co-founder and CEO of [Valuren](https://www.linkedin.com/company/valuren/), which gives fashion and luxury brands digital product passports. I work on verification for AI: checking claims against evidence, evaluating LLM systems, and planning safely in learned latent spaces.
The page at the top of this Space pulls live download numbers from the Hub API.
## VeriSci: scientific claim verification
<a href="https://huggingface.co/spaces/rishhh/verisci-claim-space"><img alt="Open the live demo in Hugging Face Spaces" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/open-in-hf-spaces-sm.svg"></a>
<a href="https://huggingface.co/rishhh/verisci-claim-verifier-retrieval-adapted-seed123"><img alt="Release verifier downloads, live" src="https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fhuggingface.co%2Fapi%2Fmodels%2Frishhh%2Fverisci-claim-verifier-retrieval-adapted-seed123%3Fexpand%255B%255D%3DdownloadsAllTime&query=%24.downloadsAllTime&label=verifier%20downloads&logo=huggingface&style=flat-square&color=2a78d6"></a>
<a href="https://huggingface.co/rishhh/verisci-scifact-e5-retriever"><img alt="Retriever downloads, live" src="https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fhuggingface.co%2Fapi%2Fmodels%2Frishhh%2Fverisci-scifact-e5-retriever%3Fexpand%255B%255D%3DdownloadsAllTime&query=%24.downloadsAllTime&label=retriever%20downloads&logo=huggingface&style=flat-square&color=2a78d6"></a>
Paste a scientific claim and get SUPPORTS, REFUTES or NOT ENOUGH INFO, with the sentences that justify the label.
<img alt="VeriSci pipeline: claim, retrieve with BM25 and a fine-tuned e5 retriever, select evidence sentences, verify with a calibrated DeBERTa model, return a label with evidence. Retriever Recall@5 0.867. Verifier accuracy 0.905, macro-F1 0.874, calibration error 0.024. End to end on SciFact validation: 0.691 accuracy, 0.653 macro-F1." src="https://huggingface.co/spaces/rishhh/about/resolve/main/assets/verisci-auto.svg" width="100%">
| Piece | Repo |
| :-- | :-- |
| Live demo | [verisci-claim-space](https://huggingface.co/spaces/rishhh/verisci-claim-space) (may take a minute to wake if idle) |
| Release verifier | [verisci-claim-verifier-retrieval-adapted-seed123](https://huggingface.co/rishhh/verisci-claim-verifier-retrieval-adapted-seed123) |
| Retriever | [verisci-scifact-e5-retriever](https://huggingface.co/rishhh/verisci-scifact-e5-retriever) |
| Ablations | Seeded variants of evidence selection (joint, evidence-gated, hard-negative, contradiction-adapted), each with its own model card |
## SchemaSage-SQL: safe text-to-SQL
<a href="https://huggingface.co/rishhh/schemasage-sql-qwen3-4b-clean-balanced-200"><img alt="Model on Hugging Face" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/model-on-hf-sm.svg"></a>
<a href="https://huggingface.co/datasets/rishhh/schemasage-sql-clean-text2sql"><img alt="Dataset on Hugging Face" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/dataset-on-hf-sm.svg"></a>
<a href="https://huggingface.co/rishhh/schemasage-sql-qwen3-4b-clean-balanced-200"><img alt="Release model downloads, live" src="https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fhuggingface.co%2Fapi%2Fmodels%2Frishhh%2Fschemasage-sql-qwen3-4b-clean-balanced-200%3Fexpand%255B%255D%3DdownloadsAllTime&query=%24.downloadsAllTime&label=model%20downloads&logo=huggingface&style=flat-square&color=2a78d6"></a>
<a href="https://huggingface.co/datasets/rishhh/schemasage-sql-clean-text2sql"><img alt="Dataset downloads, live" src="https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fhuggingface.co%2Fapi%2Fdatasets%2Frishhh%2Fschemasage-sql-clean-text2sql%3Fexpand%255B%255D%3DdownloadsAllTime&query=%24.downloadsAllTime&label=dataset%20downloads&logo=huggingface&style=flat-square&color=2a78d6"></a>
QLoRA adapters on Qwen3-4B that write read-only SQL grounded in the schema you give them, with a safety layer that refuses destructive requests.
<img alt="SchemaSage-SQL release baseline on 64 cleaned held-out examples: 100% SQL parse validity, 98.3% schema adherence, 0% unsafe queries, 100% correct refusals. QLoRA on Qwen3-4B-Instruct-2507, trained on 111,444 cleaned examples, 10,862 of them refusals." src="https://huggingface.co/spaces/rishhh/about/resolve/main/assets/schemasage-auto.svg" width="100%">
| Piece | Repo |
| :-- | :-- |
| Release baseline | [schemasage-sql-qwen3-4b-clean-balanced-200](https://huggingface.co/rishhh/schemasage-sql-qwen3-4b-clean-balanced-200) |
| Larger candidate | [schemasage-sql-qwen3-4b-clean-balanced-8k-600-v2](https://huggingface.co/rishhh/schemasage-sql-qwen3-4b-clean-balanced-8k-600-v2): better exact match and execution accuracy, missed one blocked refusal, so not shipped |
| Dataset | [schemasage-sql-clean-text2sql](https://huggingface.co/datasets/rishhh/schemasage-sql-clean-text2sql) |
| Code | [github.com/MrRobotop/schemasage-sql](https://github.com/MrRobotop/schemasage-sql) |
## Research code
| Project | What it is | Links |
| :-- | :-- | :-- |
| [toploss](https://github.com/MrRobotop/toploss) | Five optimiser-free PyTorch regularisers that rebuild SAM-style flat-minima effects as loss penalties | <img alt="PyPI version" src="https://img.shields.io/pypi/v/toploss?style=flat-square&label=PyPI&color=2a78d6"> [Preprint](https://doi.org/10.5281/zenodo.20497841) |
| [clap-family](https://github.com/MrRobotop/clap-family) | Conservative Lapse-Action Planning for safe latent trajectory optimisation, 72 tests | <img alt="PyPI version" src="https://img.shields.io/pypi/v/clap-family?style=flat-square&label=PyPI&color=2a78d6"> [Preprint](https://doi.org/10.5281/zenodo.20467271) |
| [micro-world-model](https://github.com/MrRobotop/micro-world-model) | Hierarchical JEPA world model that plans in latent space without reconstruction or reward | [Preprint](https://doi.org/10.5281/zenodo.20480620) |
| [ariadne-eval](https://github.com/MrRobotop/ariadne-eval) | Tracing and scoring for multi-step LLM agents, 285 tests | <img alt="Tests, live status" src="https://img.shields.io/github/actions/workflow/status/MrRobotop/ariadne-eval/tests.yml?style=flat-square&label=tests"> |
| [evalforge](https://github.com/MrRobotop/evalforge) | LLM evaluation with bootstrapped confidence intervals and a CI regression gate, 538 tests | <img alt="Tests, live status" src="https://img.shields.io/github/actions/workflow/status/MrRobotop/evalforge/tests.yml?style=flat-square&label=tests"> |
## Background
MSc Artificial Intelligence, University of St Andrews (dissertation on dependent types for machine learning, supervised by Dr Edwin Brady). BSc Artificial Intelligence, VU Amsterdam.
[GitHub](https://github.com/MrRobotop) 路 [LinkedIn](https://www.linkedin.com/in/rishabh-ashok-patil/) 路 [ORCID](https://orcid.org/0009-0007-0868-9673) 路 [Website](https://www.rishabhpatil.com) 路 [Email](mailto:rishabh.a.patil@outlook.com)
|