YK
Currently building AI systems @ Gauntlet AI
Namaste(); I'm
Y a s h   K u c h e r i y a .
d r e a m
b u i l d
c r a f t
a n d
a c h i e v e .

Software Developer — Backend, AI/ML & Data. Building scalable systems with Python, Java, AWS & Spring Boot. MS Computer Science @ ASU.

Let's Talk! LinkedIn GitHub

About Me

Software Developer with 3 years of experience spanning backend engineering, AI engineering, and data science/analytics. Strong expertise in Python and Java, with hands-on experience building scalable backend systems using Spring Boot, RESTful APIs, and cloud-native architectures on AWS. Experienced in developing AI and data-driven applications, productionizing ML models, and leveraging AI-assisted development tools to accelerate delivery. Adept at working across the full lifecycle from system design to deployment and optimization in a fast-paced, collaborative environment.

0+ Years Experience
0+ Projects
0 Live Demos
Yash Vijay Kucheriya

Skills

Tech Stack

  • Languages
  • PYTHON
  • JAVA
  • JAVASCRIPT
  • TYPESCRIPT
  • Frameworks
  • REACT
  • ANGULARJS
  • SPRING BOOT
  • POSTGRESQL
  • REACT NATIVE
  • Cloud & DevOps
  • AWS
  • AZURE
  • TERRAFORM
  • ANSIBLE
  • DOCKER
  • KUBERNETES
  • ML & Data
  • PYTORCH
  • TENSORFLOW
  • KAFKA
  • GO
  • Tools
  • GIT
  • GITHUB

Education

Arizona State University

Master of Science in Computer Science
Jan 2024 – Dec 2025 | Tempe, AZ, USA

• GPA: 3.8/4.0
• Coursework: Statistical ML, Algorithms, Cloud Computing, NLP, Data Mining, Networks
• Focus on AI/ML and scalable cloud architectures

Walchand Institute of Technology

B.E. in Computer Science and Engineering
Jun 2016 – May 2020 | Solapur, India

• GPA: 9.23/10
• Coursework: Data Structures, Operating Systems, Big Data, Web Development
• Strong foundation in computer science fundamentals

Experience

Gauntlet AI

AI Engineer
Feb 2026 – Present | Austin, TX, USA

• Rigorous 10-week intensive transforming experienced engineers into builders of advanced AI systems
• Deep hands-on development with AI-accelerated tools, frameworks, and real production challenges
• Mastering full-stack AI workflows, end-to-end agentic development, and scalable model integrations
• Emphasis on rapid iteration, real software delivery under real constraints, and preparation for high-impact AI engineering roles

ZenZiee (YesTech Corp.)

Software Engineer
Jun – Dec 2025 | NJ, USA

• Built ZenZiee, a gamified social app with real-time APIs, analytics dashboards, and user sentiment tracking for 100+ beta users
• Deployed containerized microservices using AWS ECS/ECR, ALB, CloudWatch, achieving 99.9% uptime
• Designed event-driven pipelines with DynamoDB Streams, Lambda, S3, reducing processing latency by 60%; collaborated with ML/frontend teams to embed emotion classification, increasing engagement by 25%

Buildoors Lab Ltd

Data Scientist
Jan – Dec 2023 | London, UK

• Developed blockchain-based fraud detection system, reducing risk exposure by 35%
• Automated ETL pipelines and built compliance dashboards using Pandas, SQL, Tableau, Power BI
• Deployed ML models on AWS SageMaker with automated retraining and monitoring workflows

Infosys Ltd

Software Engineer
Nov 2020 – Jul 2022 | Pune, India

• Designed Java Spring Boot microservices and REST APIs for enterprise billing systems, reducing manual effort by 50%
• Improved backend performance by 35% through MSSQL query optimization
• Managed 1M+ EDI transactions per month using IBM Sterling; implemented CI/CD pipelines reducing release cycles by 45%

Projects

🔍
How does DGEMM handle matrix…
SUBROUTINEDGEMM.f : 87
FUNCTIONLSAME.f : 12
EXPLAINMatrix multiply…
Live Preview
Mar 2026
LegacyLens

RAG-Powered Legacy Code Navigator

Semantic search system enabling natural language queries against 150K+ lines of LAPACK Fortran using vector similarity and LLM analysis. 8 analysis modes — explain, dependency mapping, impact analysis, documentation generation, bug search, and more. Syntax-aware Fortran chunking with ChromaDB, Voyage-code-3 embeddings, and Gemini 2.5 Flash via OpenRouter. Interactive dependency graphs with vis-network. Sub-3s answer generation at ~$0.001/query.

FastAPI ChromaDB LangChain Gemini Python Docker
github redirect button Live Demo
$124,580Portfolio
+12.4%Return
25Holdings
What's my best performer this month?
Live Preview
Feb – Mar 2026
GhostGuide

AI-Powered Wealth Management Agent

Built an AI agent layer on top of Ghostfolio (open-source wealth management) with natural-language portfolio insights. 25 specialized tools covering portfolio analysis, performance metrics, market data, and compliance checks. Multi-layer hallucination verification system with LLM output validation. NestJS + Angular full-stack with PostgreSQL, Prisma ORM, Redis caching, and LangSmith tracing. Pre-seeded with 200+ test transactions.

NestJS Angular PostgreSQL Redis OpenRouter Docker
github redirect button Live Demo
T□○/
Sprint Goals
API Design
Deploy
Live Preview
Feb 2026 – Present
CollabBoard

Real-Time Collaborative Whiteboard with AI Assistant

Built a real-time collaborative whiteboard with infinite canvas, live multiplayer presence tracking, and cursor synchronization. Implemented state sync using Supabase Realtime for low-latency multi-user editing. AI-powered assistant generates structured outputs — retrospectives, onboarding plans, and interview simulations. Designed auth, board sharing, and role-based access with Supabase Auth + RLS.

Next.js TypeScript Supabase Docker Vercel AI
github redirect button Live Demo
LIVE EventIQ Pipeline 1,247 events/s
Ingest
Validate
AI
Store
anomaly_detected 2ms
trend_analysis 5ms
event_classified 1ms
View on GitHub
Jul 2025 – Jan 2026
EventIQ

AI-Powered Real-Time Event Intelligence Platform

Built a distributed event-processing system with independent microservices for ingestion, validation, AI analysis, and analytics. Implemented Kafka-based streaming (Redpanda) for scalable real-time processing. AI analysis service leverages GPT-4 for anomaly detection, trend analysis, and event summarization. Integrated PostgreSQL + Redis caching with rate limiting, dead-letter queues, and fault-tolerant workflows.

Kafka Next.js GPT-4 PostgreSQL Redis Docker Terraform
github redirect button
BIU AI Assist
Style: More concise
Live Preview
Jan – Feb 2026
LexivoAI

AI-Powered Writing Assistant & Content Creation Platform

Built a full-stack AI writing platform with real-time grammar checking, style suggestions, and Instagram carousel generation. Integrated OpenAI API for intelligent content templates, brand voice analysis, and writing performance analytics. Designed multi-format document management with export to PDF, images, and ZIP. Auth via Clerk, data layer with Supabase (PostgreSQL), deployed on Vercel.

Next.js TypeScript OpenAI Supabase Clerk Tailwind
github redirect button Live Demo
Aug – Dec 2024
Cloud-Based Face Recognition

Multi-Stage Cloud-to-Edge Architecture on AWS

Built an IaaS system using EC2, S3, SQS, SimpleDB with custom autoscaling supporting up to 15 EC2 instances. Implemented serverless inference via containerized AWS Lambda + ECR. Designed edge-to-cloud inference with AWS IoT Greengrass enabling real-time face detection on edge devices with secure MQTT. Integrated OpenCV + facenet-pytorch for detection and recognition.

AWS EC2 Lambda Docker Flask OpenCV PyTorch
View on GitHub
Jan – May 2025
Logical Language Modeling

LLaMA Fine-Tuning for Reasoning

Continually pretrained LLaMA 3.2 (1B & 3B) on OpenWebMath with custom loss reweighting targeting logical connectors. Designed soft and hard token reweighting to improve transformer attention mechanisms. Boosted LogiQA accuracy by 28% and improved non-monotonic reasoning on LogicBench. Performed attention-head visualization to interpret model reasoning behavior.

PyTorch Hugging Face LLaMA NLP GPU
github redirect button
Nov 2025 · Anthropic Hackathon
ReadEaseAI

AI Reading Assistant for Accessible Content

Built at the Claude × Anthropic Hackathon. NLP-driven system that analyzes and rewrites complex text into clearer formats for non-native speakers and accessibility-focused audiences. Implemented AI-powered summarization using LLMs with prompt engineering for readability while preserving context. Interactive interface for real-time simplified outputs.

Claude NLP Python Prompt Engineering
Jan – May 2025
Real-Time Graph Analytics

Kafka + Neo4j + Kubernetes Pipeline

Built a scalable pipeline transforming batch graph processing into real-time streaming. Modeled NYC Yellow Taxi data as graph relationships in Neo4j, executed PageRank and BFS via Graph Data Science. Kafka Connect for continuous ingestion generating 1500+ graph relationships in real time. Orchestrated with Docker, Kubernetes (Minikube), and Helm.

Kafka Neo4j Kubernetes Docker Helm
View on GitHub
Aug – Dec 2024
Store Sales Forecasting

ML & Deep Learning Time-Series

Compared 15+ models — XGBoost, LightGBM, CatBoost, Random Forest, ARIMA, LSTM, GRU, and ensemble boosting. Extensive feature engineering: lag features, rolling averages, temporal encoding, and categorical transforms. Boosting and deep learning models significantly outperformed traditional statistical approaches across large-scale retail data.

XGBoost LSTM Python Pandas TensorFlow
github redirect button
View on GitHub
Jan – May 2024
Fake News Detection

ML & NLP Classification

End-to-end text classification using TF-IDF with Naive Bayes, Logistic Regression, SVM, Neural Networks, and XGBoost achieving ~99.7% accuracy. Text preprocessing with NLTK — tokenization, stopword removal, feature engineering. Visualization workflows analyzing model performance, boosting behavior, and loss convergence.

XGBoost NLP TF-IDF NLTK Python
github redirect button
Jan – May 2025
Income Prediction Insights

Data Visualization & Analytics

Analyzed U.S. Census data to understand socio-economic predictors of income. Designed interactive visualizations — mosaic plots, scatter plots, parallel coordinate plots. Identified key income drivers: education level, occupation type, age, and weekly hours. Generated actionable insights for marketing strategy, scholarship planning, and workforce analytics.

Python Pandas Matplotlib Data Viz
Aug – Dec 2024
House Reconfiguration Optimization

Answer Set Programming (Clingo)

Constraint-based optimization system solving the House Reconfiguration Problem using ASP and Clingo. Modeled spatial allocation constraints for items, cabinets, and rooms with declarative logic rules. Cost-minimization prioritizing reuse over expansion. Achieved optimal configurations in ~0.005 seconds; applicable to warehouse layout and resource allocation.

ASP Clingo Logic Programming Optimization
View on GitHub
Jan – May 2024
Analyzing Network RTT

FABRIC Testbed Cloud Networking

Designed controlled networking experiments across distributed nodes on FABRIC research infrastructure to measure RTT behavior. Investigated impact of topology, routing paths, and infrastructure configuration on latency. Built automated experiment workflows and visualization pipelines to interpret performance bottlenecks.

FABRIC Networking Python Cloud
github redirect button

Contact