Skip to content

CoderGuy009/CoderGuy009

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

7 Commits
Β 
Β 

Repository files navigation

Hi, I'm Aniket Sahu πŸ‘‹

πŸŽ“ Final Year B.Tech (IT) β€” Heritage Institute of Technology, Kolkata
πŸ€– AI/ML Engineer focused on LLMs, Agentic Systems, and Real-Time AI Pipelines
πŸ“ Kolkata, West Bengal


🧠 About Me

I build production-grade AI systems β€” not just demos. My work spans fine-tuning TTS models for real-time voice pipelines, designing multi-agent LangGraph workflows, and deploying low-latency FastAPI backends for conversational AI.

Currently exploring: Agentic AI Β· RAG Systems Β· LLM Fine-tuning Β· Voice AI


πŸ’Ό Experience

AI Intern β€” Hiyvr.ai (Ethical Den) Jul 2025 – Sep 2025

  • Fine-tuned Orpheus TTS model for Bengali customer support with 6-7s end-to-end latency in a real-time multi-call pipeline
  • Benchmarked open-source Bengali LLMs across accuracy, response quality, and latency for production deployment
  • Integrated Twilio and Ngrok for real-time communication, cutting setup time by 40%

Summer IS Intern β€” Indian Oil Corporation Ltd. Jun 2024 – Jul 2024

  • Built "Doc Connect", a doctor appointment management system using .NET, SQL, HTML, CSS, JS
  • Implemented role-based authentication and secure access control

πŸš€ Projects

LangGraph Β· LangChain Β· FAISS Β· Ollama (Llama 3) Β· sentence-transformers

  • Vector-based persona routing using FAISS and cosine similarity to match posts to AI bot personas
  • 3-node LangGraph state machine that autonomously researches topics and generates structured JSON posts via RAG
  • Prompt injection defense using role-separated system prompts and identity anchoring

FastAPI Β· LangGraph Β· OpenAI API Β· Python

  • Multi-agent conversational AI system automating booking, cancellation, and query workflows β€” 70% reduction in manual intervention
  • FastAPI backend handling real-time requests with < 1s latency
  • Automated slot validation pipeline improving scheduling accuracy by 85%

FastAPI Β· PyTorch Β· Transformers

  • Transformer-based models for accurate medical text extraction and inference from PDF documents
  • Structured data extraction pipeline converting unstructured PDFs into actionable insights

πŸ› οΈ Tech Stack

Languages: Python Β· C Β· SQL
AI/ML: LLMs Β· RAG Β· NLP Β· Deep Learning Β· Transformers Β· LoRA Fine-tuning Β· Vector Databases (FAISS)
Frameworks: LangChain Β· LangGraph Β· PyTorch Β· TensorFlow Β· Scikit-learn Β· FastAPI
Tools: Ollama Β· HuggingFace Β· Docker Β· Git Β· GitHub
Data: Tableau Β· Data Analysis


πŸ“Š GitHub Stats

Aniket's GitHub stats Top Langs


🌐 Connect

LinkedIn Email GitHub

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors