Hi, I'm

Samarth Patel

>_

AI & ML B.Tech student at Symbiosis Institute of Technology, Pune, with hands-on experience building production AI agents, RAG systems, and full-stack platforms. Currently interning at Futurecraft, deploying AI microservices on AWS EKS and Kubernetes. Passionate about applied AI — from LLM pipelines to MLOps infrastructure.

View ProjectsDownload Resume

About Me

Building at the intersection of AI and software engineering.

AI & ML B.Tech student at Symbiosis Institute of Technology, Pune, with hands-on experience building production AI agents, RAG systems, and full-stack platforms. Currently interning at Futurecraft, deploying AI microservices on AWS EKS and Kubernetes. Passionate about applied AI — from LLM pipelines to MLOps infrastructure.

Pune, Maharashtra, India

Bachelor of Technology — Artificial Intelligence & Machine Learning

Symbiosis Institute of Technology, Pune

Aug 2022 – PresentCGPA: 7.25

22+

Public Repos

1

npm Package

6+

Hackathons

AI @ FC

Internship

Experience

AI Intern

Futurecraft

Dec 2025 – Present
  • ▸Built production-ready AI agents for the Core Asset Tokenization Platform — Document Parsing (AWS Textract OCR), Document Generation (Jinja2 legal engine with blockchain hash recording), and Reasonableness Check (valuation QA with confidence scoring and anomaly detection).
  • ▸Integrated DSPy into the GYCO AI Chatbot, replacing manual prompts with declarative optimisable pipelines using ChainOfThought signatures for KPI summarisation and trend analysis.
  • ▸Deployed and orchestrated AI agent microservices on AWS EKS and Kubernetes; configured IAM, VPC networking, auto-scaling, and performed distributed pod log analysis for production debugging.
PythonFastAPIDSPyAWS TextractAWS EKSKubernetesDockerPostgreSQLGitLab CI/CDQwen3 14B

Skills

Technologies and tools I work with across the AI/ML and full-stack ecosystem.

Languages

PythonTypeScriptJavaScriptJavaCHTML / CSS

AI / ML

LLMs & Prompt EngineeringRAG SystemsDSPyLangChainGoogle GeminiFAISSXGBoost / LightGBMscikit-learnTensorFlow / KerasOpenCVNLPComputer Vision

Frontend

Next.js 15ReactTailwind CSSFramer MotionRecharts / Plotly.js

Backend & APIs

FastAPIFlaskExpress.jsNode.jsREST APIsOpenAPI / SwaggerNextAuth.jsPydantic

DevOps & Cloud

DockerKubernetesAWS EKSAWS S3 / EC2AWS TextractGitHub ActionsGitLab CI/CDDVC

Databases

PostgreSQLMongoDBFAISS Vector DBSQL

Projects

A selection of things I've built — from AI systems to full-stack platforms.

Retirewise — AI Pension Platform

Featured

Full-stack pension management and retirement planning platform with AI chatbot, Monte Carlo simulations, and role-based dashboards for Members, Advisors, and Regulators.

  • ▸AI-powered chatbot using locally integrated LLaVA vision LLM for retirement guidance
  • ▸Monte Carlo simulations and interactive what-if scenario analysis
  • ▸Role-based dashboards with JWT authentication via NextAuth.js
Next.js 15Express.jsPostgreSQLLLaVATailwind CSS+2 more
Code

claude-log-analyzer

Featured

MCP server that integrates with Kubernetes clusters to give Claude Code deep log analysis, anomaly detection, and root cause analysis capabilities. Published on npm.

  • ▸Fetches pod logs, status, metrics, and events from live K8s clusters
  • ▸Anomaly detection: CPU/memory spikes, error bursts, restart storms, OOM kills
  • ▸Evidence-backed root cause analysis reports in Markdown or TXT
TypeScriptKubernetesMCPNode.js
Codenpm

LLM-Powered Intelligent Query Retrieval System

Featured

Retrieval-Augmented Generation (RAG) document intelligence system built for Bajaj HackRX. Processes PDFs and delivers contextually accurate answers via vector embeddings and FAISS.

  • ▸PDF ingestion with adaptive text chunking preserving context
  • ▸High-quality vector embeddings using FlagEmbedding models
  • ▸Fast similarity search via FAISS indexing
PythonFastAPIGoogle GeminiFAISSFlagEmbedding+1 more
Code

Traffic Volume MLOps Pipeline

Featured

Production-grade MLOps pipeline for predicting highway traffic volume using weather and time-series data. Features DVC orchestration, multi-model ensemble, and AWS cloud deployment.

  • ▸End-to-end ML pipeline with DVC orchestration and data versioning
  • ▸Ensemble of XGBoost, Random Forest, LightGBM — R² ~0.85–0.90
  • ▸Flask web app with user retrain capability
PythonDVCXGBoostLightGBMFlask+3 more
Code

AnimeGAN with Emotion Control

GAN-based model that transforms realistic human faces into anime-styled characters with intensity-based emotion control using e4e Encoder and StyleGAN2 Decoder.

  • ▸Transforms human faces into anime-style characters
  • ▸e4e Encoder + StyleGAN2 Decoder for intensity-based emotion control
  • ▸Deep learning computer vision with PyTorch
PythonPyTorchStyleGAN2AnimeGANv2Torchvision
Code

RAG LLM Conversational App

Conversational RAG web app for chatting with documents, websites, and custom data sources. Built with Streamlit for rapid prototyping and LangChain for LLM orchestration.

  • ▸Chat with documents, websites, and custom data sources
  • ▸LangChain for LLM orchestration
  • ▸Rapid prototyping with Streamlit UI
PythonStreamlitLangChainRAG
Code

Achievements

Hackathons competed in and certifications earned.

Hackathons

Bajaj HackRX 6.0

Bajaj Finserv

Top 80

AmazonML Challenge

Amazon

Top 100

Datathon by CCOEW

CCOEW

Top 20

TensorCraft

Finalist

ByteVerse 7.0

MUFG Hackathon

MUFG

Participant — Built Retirewise

Certifications

Python 101 for Data Science

IBM

Big Data 101

IBM

Get In Touch

Whether it's a project, opportunity, or just a hello — I'd love to hear from you.

Based in Pune, India (UTC +05:30). Usually responds within 24 hours.