Open to opportunities

Ankit Singh

@ankitsingh39

Message

I build production-ready AI systems using LLMs, RAG pipelines, agent orchestration, and Python.

India

Message

What I'm looking for

I’m looking to build and deploy production-ready AI/ML solutions, especially language-model, agentic AI, and RAG systems using Python, deep learning, and NLP.

I build AI/ML applications around language models, Retrieval-Augmented Generation, and agentic workflows. My work combines Python, LangChain, LangGraph, FastAPI, Streamlit, and production-oriented deployment.

I built an autonomous research agent with LangGraph that uses live web search, calculation, memory, retry-with-backoff, and step budgets to support reliable multi-turn research. I deployed it as a FastAPI REST service and Streamlit application with defensive tool-boundary error handling.

I’ve also built a document Q&A chatbot grounded in vector-database retrieval and fine-tuned BERT/DistilBERT models with PEFT/LoRA for domain-specific classification. My work is supported by IEEE publications spanning NLP, depression detection from social media text, and real-time anomaly detection.

Experience

Work history, roles, and key accomplishments

Autonomous Research Agent

Personal Project

Jul 2026 - Sep 2026 (2 months)

Built a multi-step autonomous agent using LangGraph with tool calling, retry-with-backoff, and a step budget. Deployed as a FastAPI REST service and a live Streamlit application.

Python LangGraph fastAPI Streamlit Tool calling Agent orchestration

RAG-Based AI Chatbot Developer

Personal Project

Apr 2026 - Jun 2026 (2 months)

Built a Retrieval-Augmented Generation pipeline that indexes documents into a vector database and grounds LLM responses in retrieved context. Deployed via FastAPI backend with a chat interface.

Python OpenAI API RAG Generative AI fastAPI Vector Databases Semantic Search

Fine-Tuning Transformer Model

Personal Project

Jan 2026 - Mar 2026 (2 months)

Fine-tuned a pre-trained transformer (BERT/DistilBERT) using Hugging Face Transformers and PEFT (LoRA) for domain-specific text classification. Built data preprocessing, tokenization, and training pipeline.

Python Hugging Face Transformers PEFT LoRA PyTorch scikit learn Model Evaluation