ayush_singh

@Ayush_SIngh

Ayush Singh

Building open source infrastructure to detect LLM hallucinations, prompt attacks...
India github.com/AyushSingh110 Joined May 2026
1.3k Points26 Badges8 Connections7 Followers8 Following

About

AI and data science undergrad student exploring new technologies and doing research on the models to make them more reliable and to make sure that there is no wrong output get deli... Show more

Top Skills

pythonMLLLMMongodbmodel securitydeep learningfastapiflaskModel developmentreactpytorch

Education

Pimpri Chinchwad University

BTech, CS(AI and Data Science)

Pursuing my bachelor degree in computer science with the specialisation in AI and data science .

Projects

Failure-Intelligence-System

Visit Project open_in_new

Open source guardrail for production LLMs. Detects prompt attacks, diagnoses hallucinations, and corrects model failures in real time before the output ever reaches your users. Ships as a Python SDK with a single decorator.

Licenses & Certifications

Generative AI Professional

Oracle
Show Credential open_in_new

AI Foundation Associate

Oracle
Show Credential open_in_new

Complete Data Science, Machine Learning, DL, NLP Bootcamp

Udemy
Credential ID UC-49f5f75a-c1bb-436b-999f-ac171daaaea6
Show Credential open_in_new

Language & Tools

Python , FastAPI , Flask , Mongodb , Model finetuning, Machine learning, RAG, Langchain , Langraph

Currently Exploring

Exploring real-time LLM reliability. Building a system that monitors, diagnoses, and intercepts model failures - hallucinations, prompt attacks, and adversarial inputs before the output ever reaches the user.

Achievements

Published open source LLM safety SDK (fie-sdk) on PyPI real-time hallucination and prompt attack detection used by developers building production AI systems.
Got 2500+ downloads on the package in just 40 days.

Got certification in generative ai (professional) and ai associate by oracle

Fun Fact

I got curious about why LLMs fail silently so I built an entire open source monitoring system to find out. What started as a side project is now a PyPI package used by developers across the world.

Random Dev Quote

"Don't just ship the model. Ship the system that watches it"
Joined: 3 weeks (since May 10)
Extra privileges: Editing any comment
Full Name: Ayush Singh
Headline: Building open source infrastructure to detect LLM hallucinations, prompt attacks, and model failures in real time.
About: AI and data science undergrad student exploring new technologies and doing research on the models to make them more reliable and to make sure that there is no wrong output get deliever to the user from model. Working on different attacks which are been done on the model and how to protect them in real time.
Location: India
Website: https://github.com/AyushSingh110
Languges & Tools: Python , FastAPI , Flask , Mongodb , Model finetuning, Machine learning, RAG, Langchain , Langraph
Currently Exploring: Exploring real-time LLM reliability. Building a system that monitors, diagnoses, and intercepts model failures  - hallucinations, prompt attacks, and adversarial inputs  before the output ever reaches the user.
Achievements: Published open source LLM safety SDK (fie-sdk) on PyPI  real-time hallucination and prompt attack detection used by developers building production AI systems.
Got 2500+ downloads on the package in just 40  days.

Got certification in generative ai (professional) and ai associate by oracle
Fun Fact: I got curious about why LLMs fail silently  so I built an entire open source monitoring system to find out. What started as a side project is now a PyPI package used by developers across the world.
Random Dev Quote: "Don't just ship the model. Ship the system that watches it"
Top Tags

Top Tags

Activity

User Activities

JanFebMarAprMayJunJulAugSepOctNovDec
Mon
Tue
Wed
Thu
Fri
Sat
Sun
Less More
Contribution
chevron_left

Latest Jobs

View all jobs →

Active in these Groups:

AI
1023 members
Dev Tools
582 members
Open Source
966 members