AI Engineer & Builder

Smriti
Asthana

AI Engineer  ·  NLP  ·  RAG  ·  Backend Engineering

Practical AI systems, built from model to deployment.

I'm a Computer Science and Engineering (Artificial Intelligence) graduate building document-intelligence systems, NLP applications, production APIs, and cloud-deployed machine-learning solutions.

View Featured Work Download Résumé GitHub Contact Me
Smriti Asthana, AI Engineer
About

Building complete AI systems

My work sits at the intersection of machine learning, language technology, backend engineering, and practical product development. I build end-to-end systems involving document ingestion, data processing, embeddings, retrieval, model integration, APIs, evaluation, interfaces, and cloud deployment.

I co-founded TEAM-PSMPV and lead technology delivery across websites, Android applications, AI agents, chatbots, automation solutions, and managed IT services.

My focus is on shipping systems that work reliably in production — not just notebooks that demonstrate a concept. I care about evaluation, hallucination handling, latency, and making AI tools genuinely useful to the people who use them.

Areas of focus

  • Machine Learning & NLP
  • Retrieval-Augmented Generation
  • Backend Engineering & FastAPI
  • API Development & Design
  • Cloud Deployment (AWS · Azure)
  • AI Evaluation & Testing
  • Document Intelligence
  • Practical Product Development
Featured Work

Selected projects

Android Application

OfflineTTS — Privacy-First Offline Text-to-Speech

View on Google Play

OfflineTTS is a privacy-first Android text-to-speech application that operates completely offline after the selected voice is downloaded. It requires no login, subscription, API key, or continuous internet connection.

The application cleans and divides long text into manageable chunks before processing it through Kokoro or Piper voice models. The models perform grapheme-to-phoneme conversion and synthesize natural speech directly on the user's device.

Key features

  • Fully offline text-to-speech
  • Eight supported languages
  • 70,000+ character support
  • Audio export
  • No login or subscription
  • No API-usage limits
  • On-device inference
  • Text cleaning & chunking

Technology

Kotlin ONNX Runtime Kokoro TTS Piper TTS Text Chunking Android

AI · RAG · Backend

Research Assistant — RAG-Based Document Question Answering

GitHub Repository Live Application

A retrieval-augmented generation system that allows users to upload one or more PDF documents and ask natural-language questions about their contents. The system provides accurate, source-attributed answers with confidence scoring and conversational memory.

Key features

  • PDF upload & text extraction
  • Smart text chunking
  • Sentence Transformer embeddings
  • FAISS semantic retrieval
  • Multi-document search
  • Gemini-powered generation
  • Conversational memory
  • Source attribution
  • Confidence scoring
  • Chat export
  • FastAPI backend
  • Streamlit frontend

Technology

Python FastAPI Streamlit FAISS Sentence Transformers Gemini API Docker
Projects

Selected Repositories

Research Assistant

End-to-end RAG system for PDF question answering — FastAPI backend, FAISS retrieval, Sentence Transformer embeddings, Gemini answer generation, and conversational memory with source attribution.

Python FastAPI FAISS RAG Gemini API
GitHub Live App

Fake Review Detection

An end-to-end NLP classification system using TF-IDF, Logistic Regression, SVM, XGBoost, and DistilBERT. The strongest models achieved approximately 88% accuracy, and the application was deployed on Azure App Service using GitHub Actions CI/CD.

Python DistilBERT XGBoost Scikit-learn Azure
GitHub Hugging Face Live Demo

Flight Price Prediction Using AWS SageMaker

An end-to-end XGBoost regression workflow covering data cleaning, exploratory analysis, feature engineering, AWS SageMaker experimentation, Amazon S3 storage, model evaluation, and an interactive Streamlit prediction interface.

Python XGBoost AWS SageMaker Amazon S3 Streamlit
GitHub

FastAPI Patient Management System

A production-inspired REST API with patient CRUD operations, automatic BMI and health-verdict generation, PostgreSQL persistence, SQLAlchemy ORM, Redis caching, Celery background tasks, Alembic migrations, and Docker Compose.

Python FastAPI PostgreSQL Redis Docker
GitHub
View All GitHub Repositories →
Experience

Work history

TEAM-PSMPV

June 2026 – Present

Remote

Founding Team Member & Team Lead

  • Co-founded TEAM-PSMPV and lead technology delivery across websites, Android applications, AI agents, chatbots, automation solutions, and managed IT services.
  • Launched OfflineTTS on Google Play and delivered a responsive website for Anand Hospital.
Handshake AI

July 2026 – August 2026

Remote

AI Trainer — Project Dynamo

  • Created and evaluated command-line benchmark tasks to assess AI agents on multi-step terminal workflows, debugging, and verifiable output generation.
WoneGlobal

June 2025 – July 2025

Onsite · Delhi

Founding Office Intern

  • Automated routine workflows and collaborated across teams to streamline operations, reducing manual effort by approximately 15%.
Publication

Research

Peer-reviewed · IJERCSE

Fake Review Detection in E-Commerce: A Critical Review of Robustness, Generalization, and Deployment Challenges

IJERCSE, Volume 13, Issue 05, May 2026

A review paper analysing machine-learning, NLP, and transformer-based approaches to fake-review detection, including robustness, generalization, and deployment challenges.

Read Publication
Technical Skills

Capabilities

Programming

Python SQL JavaScript HTML CSS

Machine Learning & NLP

PyTorch Scikit-learn Transformers Hugging Face TensorFlow NLTK XGBoost TF-IDF

Generative AI

RAG LLMs FAISS Sentence Transformers Gemini API LangChain Prompt Engineering

Backend & Databases

FastAPI REST APIs SQLAlchemy Pydantic PostgreSQL Redis Celery Alembic

Cloud & DevOps

AWS SageMaker Amazon S3 Azure App Service Docker Git GitHub GitHub Actions

Data Analysis

NumPy Pandas Matplotlib Feature Engineering Model Evaluation
Education

Academic background

Nov 2022 – Jun 2026

Galgotias College of Engineering and Technology

B.Tech in Computer Science and Engineering (Artificial Intelligence)

CGPA: 7.3

Class XII · CBSE

Senior Secondary

89.9%

Class X · CBSE

Secondary

94.4%

Certifications

Credentials

Oracle Cloud Infrastructure Certified AI Foundations Associate

Oracle · October 2025 · Credential: 103012803OCI25AICFA

View Credential

AWS Academy Graduate — Machine Learning for Natural Language Processing

AWS Academy · April 2026

View Credential

Supervised Machine Learning: Regression and Classification

Coursera · Andrew Ng · August 2025

View Credential

McKinsey Forward Program

McKinsey & Company · June 2026

View Credential
Writing

Technical articles

01

My First Steps with FastAPI: Building a Patient Management REST API

A walkthrough of building a production-structured REST API with FastAPI — covering routing, Pydantic models, and data persistence.

Read Article

02

Building a Production-Ready FastAPI Backend with PostgreSQL, Redis, Celery & Docker

A deep dive into constructing a scalable backend stack — integrating async task queues, caching, database migrations, and containerisation.

Read Article
View All Articles on Hashnode →

Let's build something useful.

I'm open to AI/ML, NLP, RAG, backend, and cloud-engineering opportunities, as well as meaningful technical collaborations.

Email

asthanaa.smriti96@gmail.com

GitHub

github.com/Smriti2929

LinkedIn

linkedin.com/in/smriti-asthana-3518bb315

Hashnode

hashnode.com/@smriti29

Résumé

Download PDF