Graduated May 2026 · OPT June 2026

Hi, I'm Rigved. I build

MS in Computer Science from CU Boulder · IEEE-published in computer vision · Recently shipped RAG infrastructure at Rule4 and multimodal GenAI products at The Builder Market. Turning research into systems that ship.

Ask my resume Download resume

github.com/rigvedthorat linkedin.com/in/rigved-thorat rigvedthorat@gmail.com

scroll

Ask my resume anything

A static RAG demo over my resume and portfolio facts. It ranks embedded snippets in your browser, then assembles a concise recruiter-style answer with the evidence it used.

0 API keys or backend calls

11 resume chunks indexed client-side

About

I'm an AI/ML Engineer who likes the messy middle — where research papers become production systems, and "we should try an LLM here" turns into a working RAG pipeline that's actually fast, scoped, and observable.

As my CU Boulder capstone at Rule4, I built managed RAG infrastructure on AWS Bedrock and OpenSearch — an AI ticket-drafting feature that retrieves client-scoped historical context and orchestrates prompts under a 6-second P95 budget. Earlier, at The Builder Market, I shipped multimodal GenAI tools spanning image generation, smart interior editing, and vision-based home repair analysis.

Before that, I co-authored an IEEE conference paper on CNN-based plant disease classification (95.8% accuracy on 10 classes). I care about evaluations more than demos, latency budgets more than benchmarks, and shipping more than slideware.

3.70

GPA · MS CS
CU Boulder

20×

Precision@5 lift
over zero-shot baseline

95.8%

Accuracy on IEEE
CNN paper

P95 end-to-end
RAG latency shipped

Experience

AI Engineer Capstone

Rule4, Inc. · Boulder, CO · Hybrid

Oct 2025 — Apr 2026

Built an AI-assisted ticket drafting feature on Rule4's TFox helpdesk platform that generates context-aware titles & descriptions from user input and attachments — Python, Django, AWS Bedrock, OpenSearch.
Designed end-to-end RAG pipelines using AWS Bedrock Titan embeddings and OpenSearch kNN search to retrieve top-5 client- and queue-scoped similar tickets for prompt orchestration.
Developed backfill and on-save ETL workflows that vectorize historical TFox tickets from MySQL into an OpenSearch index, lifting draft relevance and semantic search across past tickets.
Shipped feedback & analytics dashboards tracking retrieval behavior and P95 latency across embedding, search, and generation stages — end-to-end latency of 6 seconds.

PythonDjangoAWS BedrockOpenSearchMySQLRAG

AI Product & Prototyping Intern

The Builder Market · Los Angeles, CA · Remote

Jun 2025 — Sep 2025

Built Angular image generation workflows backed by NestJS REST APIs, integrating Azure OpenAI GPT Image 1 and Black Forest Labs Flux for text- and image-based home design generation.
Engineered 3 smart interior editing tools for paint, object replacement, and surface reskinning using GPT-4o image analysis, reference-image inputs, and Sharp-based masking, bounding boxes, and feathering logic.
Developed a multimodal DIY repair analysis flow with GPT-4o Vision converting uploaded home images into 4-part repair guidance: tools, materials, repair steps, and cost/time estimates.

AngularNestJSTypeScriptAzure OpenAIGPT-4o VisionSharp

Selected Projects

Featured · Research Apr — May 2025

LLMs as Movie Recommendation Systems

Fine-tuned a GPT-4o-mini model on the Netflix Prize dataset (50K+ viewers, 18K+ movies) to build a recommender that understands complex viewer preferences. Constructed custom training data by sampling 10K viewers and identifying "ground truth" movies rated 5★ by >50% of similar groups.

20× recommendation relevance vs. baseline

0.02 → 0.42 Precision@5 lift

PythonGPT-4o-miniFine-tuningPyTorchPandas

IEEE Published Nov 2023 — Mar 2024

Image-Based Tomato Leaf Disease Classification

First-authored an IEEE conference paper on a CNN for multiclass tomato leaf disease classification, benchmarked against SVM, ResNet, MobileNet, DenseNet, and InceptionV3. Trained a 4-layer CNN (16/32/64/128 filters, ReLU, global average pooling) on PlantVillage with OpenCV preprocessing & augmentation.

95.8% accuracy · 10-class classification

DOI: 10.1109/INCOS59338.2024.10527655 ↗

PythonTensorFlowCNNOpenCV

Full-Stack · GenAI Mar — May 2025

Mind Bloom — AI Mental Health Journal

Full-stack mental health journal app where users log daily mood-based entries and receive personalized inspirational quotes generated from their journal context. Built a React + TypeScript frontend on a Node.js + PostgreSQL + Prisma backend, integrated the OpenAI API for mood-to-quote mapping. Deployed on GCP & Vercel.

ReactTypeScriptNode.jsPostgreSQLPrismaOpenAIGCPVercel

In Production 2025 — 2026

RAG Infrastructure @ Rule4

Managed RAG pipelines on AWS Bedrock + OpenSearch powering ticket drafting at the TFox helpdesk platform. Client-/queue-scoped kNN retrieval, on-save ETL indexing from MySQL, and observability across embedding/search/generation stages.

AWS BedrockOpenSearchDjangoVector Search

Multimodal · GenAI 2025

Smart Interior Editing Tools

Three GenAI editing tools for paint, object replacement, and surface reskinning. GPT-4o vision analysis pairs with reference-image inputs, Sharp-based masking, bounding boxes, and feathering to produce photorealistic interior edits.

GPT-4o VisionAzure OpenAIFluxSharpNestJS

Stack

◈ GenAI & LLMs

RAGPrompt EngineeringLLM Fine-tuning Vector EmbeddingsSemantic SearchMultimodal AI AgentsAWS BedrockOpenSearch Azure OpenAIGPT-4o

◆ ML / Deep Learning

PyTorchTensorFlowScikit-learn OpenCVCNNsTransfer Learning Recommendation SystemsPandasNumPy Model Evaluation

◉ Languages

PythonTypeScriptJavaScript SQLHTMLCSS

◇ Backend & Web

DjangoFlaskNode.js NestJSExpressREST APIs ReactAngular

▣ Cloud & DevOps

AWS (ECS, RDS, Bedrock, S3, Lambda)Azure AI Foundry GCPDockerGit GitHub ActionsGitLab CIVercel

▤ Data & Storage

PostgreSQLMySQLNeo4j OpenSearch (vector)Prisma ORM

Education

🇺🇸

University of Colorado Boulder

M.S. Computer Science

Aug 2024 — May 2026 GPA · 3.70 / 4.00

Boulder, CO, USA

🇮🇳

SRM University

B.Tech. Computer Science & Engineering — AI & ML

Sep 2020 — May 2024 GPA · 8.99 / 10.00

Chennai, TN, India

Let's build something

I graduated in May 2026, OPT June 2026, and I'm looking for AI/ML Engineer, Software Engineer, and Data Engineer roles across the USA. Open to hybrid, remote, or relocation.

Email rigvedthorat@gmail.com

Phone +1 (303) 435-7650

LinkedIn in/rigved-thorat

GitHub github.com/rigvedthorat

Resume