Yash Khairnar — Portfolio

About Me

The world of Machine Learning and AI has always inspired me to explore how far we can push its boundaries.

I like to research new ideas and continually challenge myself to learn and apply state-of-the-art techniques.

I hold a Bachelor's in Computer Engineering from Pune University and I'm currently pursuing an M.S. in Computer Science at San José State University.

My hands-on experience spans Machine Learning and Deep Learning, with applications in Computer Vision, NLP, and multimodal AI. Beyond AI, I'm proficient in full-stack development (MERN), cloud platforms (AWS), and orchestration tools.

Experience

Software Engineer

Accurate Industrial Controls Pvt. Ltd.

July 2024 - June 2025 | Pune, India

• Led a team of 6 in developing a remote generator monitoring system with anomaly detection, achieving 93% accuracy.
• Built an ANPR application using YOLOv11 and PaddleOCR, improving plate detection to 97% accuracy.
• Designed a ROS-based pre-filling pipeline for LPG cylinders, eliminating manual intervention.

Artificial Intelligence Intern

Accurate Industrial Controls Pvt. Ltd.

Aug 2023 - Nov 2023 | Pune, India

• Automated anomaly detection for copper coils using PatchCore and YOLO.
• Engineered a real-time video streaming pipeline with 50ms latency.
• Researched pathfinding algorithms for autonomous boat navigation.

Deep Learning Intern

ResoluteAI Software

Nov 2022 - Feb 2023 | Bengaluru, India

• Built scalable computer vision systems for face recognition and space estimation.
• Engineered OCR workflows for resume parsing, improving accuracy by 85%.
• Deployed YOLO models for industrial automation with 92%+ accuracy.

Featured Projects

Documentation Assistant

A professional Retrieval-Augmented Generation (RAG) application designed to provide AI-powered support based on technical documentation URLs. Users can index multiple documentation sources and chat with an assistant that retrieves context from the indexed material to provide accurate, technical answers.

PythonLangChainFastAPIRAG

Nook

Cafe discovery platform with personalized recommendations using collaborative filtering and location-based ML

Next.jsMachine LearningReact NativeAWSCI/CD

ThinkFlow

EEG-to-text translation system using VQ-VAE + BART Transformer and LSTM Seq2Seq with Attention for real-time brain signal decoding

PyTorchHuggingFaceTransformersFlaskNext.jsLSTMVQ-VAE

Malware Classification

Classified malware into 12 families using DenseNet121 CNN, achieving 95% accuracy on 10K+ samples

PyTorchCNNDenseNet121SVMGrid Searchencoder-decoderMachine LearningDeep Learning

Sentinel

Serverless email marketing platform built on AWS with real-time analytics, AI-powered content generation, and multi-region deployment.

AWS LambdaAWS S3AWS API GatewayDynamoDBSQSSNSSESTerraformNext.jsGitHub Actions

Agies

An AI-powered automated bug fixing agent that analyzes Sentry errors, identifies problematic code, proposes fixes, tests them in a sandbox environment, and creates draft pull requests.

Agentic AILangGraphStreamlitSentryCodeRabbitDaytone SandboxGithub API

AetherForge

It's a multi-agent AI IDE that allows multiple humans and specialized AI agents to edit the same codebase simultaneously using Gemini orchestration and Automerge (CRDTs) for conflict-free, real-time collaboration.

CrewAIAgentic WorkflowAutomergeReact.js

CuriosityAI

An AI-powered agentic system that identifies innovation gaps in research landscapes using embedding inversion and retrieval-augmented generation, autonomously producing novel invention abstracts, sketches, and proposals to democratize R&D. 20+ validated hypotheses from 10K+ embeddings

CalHacks 12.0Machine learningAgentic AIVoice AgentEmbedding InversionRAG

ResuMatch

RL-based resume optimizer using a two-agent architecture (Editor + Judge) with reward shaping to tailor resumes for specific job postings

Next.jsTavilyReinforcement LearningLangGraphGitFlask

Slique

A platform for connecting Fashion Models to Brand recruiters. Built with features like job posting, searching, application, real-time chatting, and online contract-based hiring.

Next.jsTailwindCSSFastAPIPostgreSQLAWS RDSGit

Image Colorizer

Give colors to your black and white photographs using deep learning models.

FlaskHTML/CSSNumPyOpenCvGANModel Finetuning

4Bit

A tool designed to diagnose lung cancer with the help of multi-omics data and CT scan images.

PythonKerasTensorflowMulti OmicsBiomarkersGenetic Mutations

Roomie Radar

Roommate matching app that uses preference-based compatibility scoring to suggest the best fits.

Next.jsMachine Learning

Technical Skills

Machine Learning & AI

RegressionClassificationSVMDecision TreesRandom ForestDeep LearningCNNsTransformersNLPRLGANsLoRA/PEFTRAGLangGraph

Libraries & Frameworks

TensorFlowKerasPyTorchScikit-LearnNumPyPandasReactNode.jsNext.jsFlaskLangChain

Databases

MySQLMongoDBPostgreSQLRedisDynamoDB

Cloud & DevOps

AWSEC2RDSAmplifyDockerKubernetesGitNginxDistributed Systems

Languages

PythonC++JavaScriptTypeScriptGoHTML/CSS

Achievements & Leadership

Awards

Best Student Award - IIIT

March 2024

Recognized for academic and extra-curricular excellence throughout 4 years.

Leadership

Project Lead

Led cross-functional team of 6 engineers in developing industrial ML solutions.

Chairman - CESA (I2IT)

Led technical and social programming for student development.