DINASQUARE

AI/ML Engineer & Full Stack Developer · Lucknow, India

GAURAANGIPRAAKASH

Research. Systems. Shipping.
Building production AI at the intersection of LLMs, computer vision, and full-stack — currently looking for roles where I can do all three.

scroll
Python · TypeScript · C++LLM SystemsRAG PipelinesComputer VisionTensorFlow · PyTorchFull-Stack EngineeringNext.js · React · FastAPIAWS · GCP · AzureMultimodal AINLP ArchitecturesPublished ResearcherPrompt EngineeringDeep LearningDocker · GitPython · TypeScript · C++LLM SystemsRAG PipelinesComputer VisionTensorFlow · PyTorchFull-Stack EngineeringNext.js · React · FastAPIAWS · GCP · AzureMultimodal AINLP ArchitecturesPublished ResearcherPrompt EngineeringDeep LearningDocker · Git
01Experience

WHERE I'VE
SHIPPED.

A
Engineer I
Creatium · AI/ML & Full Stack
Sep 2025 – Feb 2026
  • Multi-LLM Math Solver API — latency crushed from 30s → 6s at 86% accuracy
  • Developed Build Your Activity AI simulation builder
  • Shipped Make Your Avatar: webcam recorder with Azure Blob + GCP deployment
  • Engineered Qatari Arabic translation agent via XML prompt engineering
B
Research Intern
IIT Kanpur · Dept. of CSE
Jun 2024 – Aug 2024
  • Built ResEmoteNet emotion classifier — 88% accuracy on FER2013
  • Designed real-time inference pipeline for facial-expression analysis
C
AI Intern
YOKO Solutions
May 2024 – Jul 2024
  • Built NLP embedding-based bio-generation system for matchmaking platform
  • UI/UX design in Figma — 30% engagement improvement
02Selected Work

THINGS I'VE
BUILT.

PROJECT 001

LYREBIRD

Interactive 3D language learning tutor with real-time conversation, RAG-enhanced responses, and an immersive Three.js environment that makes learning feel like play.

Next.jsThree.jsGemini APITypeScriptClerkRAG
View on GitHub →

Click for highlights ↑

Key Highlights

LYREBIRD

3D EnvironmentReact Three Fiber with VRM & GLB avatar support
Voice I/OElevenLabs TTS + live speech-to-text recognition
AI EngineGoogle Generative AI for natural conversation
RAG PipelineSupabase pgvector — context-aware responses
Chat HistoryPersistent sessions across conversations
BackendNext.js API routes for chat, history & speech
View on GitHub →
001

PROJECT 002

MULTIMODAL SENTIMENT ANALYSIS

Full-stack web app analyzing sentiment in videos across visual, audio, and text modalities. Combines facial emotion detection, audio sentiment, and Whisper transcription into a unified confidence score.

Next.jsFastAPIPythonTensorFlowWhisperComputer VisionNLPChart.js
View on GitHub →

Click for highlights ↑

Key Highlights

MULTIMODAL SENTIMENT ANALYSIS

Visual ModalityFacial emotion detection via dima806 model
Audio ModalityTone & prosody sentiment from raw audio
Text ModalityWhisper large-v3 transcription → NLP analysis
Unified OutputCombined confidence score across all 3 modalities
Real-timeLive processing status updates during analysis
ExportFull results as a downloadable PDF report
View on GitHub →
002
03Tech Stack

WHAT I
WORK WITH.

Languages

PythonC++JavaTypeScriptSQL

AI / ML

TensorFlowPyTorchscikit-learnNumPyPandasHugging FaceLLMs & RAGComputer Vision

Web & Full-Stack

Next.jsReactNode.jsFastAPITailwind CSS

Cloud & DevOps

AWSGCPAzureDockerGitGitHub

Databases

PostgreSQLMongoDBMySQLPineconeSupabase

Currently Exploring

AgentsMultimodal LLMsLangChainFine-tuning
04Research

WHAT I'VE
PUBLISHED.

05Education

WHERE I
LEARNED.

Amity University

B.Tech · Computer Science Engineering

Aug 2021 – Jun 2025

0.00 CGPA

Sheiling House School

12th · CICSE PCMB

Apr 2020 – Jul 2021

0.00%

Sheiling House School

10th · ICSE Science

Apr 2018 – May 2019

0.00%

LET'S BUILDSOMETHING.

Open to AI/ML engineering roles, research collaborations, consulting, and interesting conversations. Based in Lucknow, India — available remotely worldwide.

— Send A Message

I'm Interested In

— Honours

Hackathon Winner

Hack to Crack 1.0 · May 2024

Content Designer

GDG Lucknow

ERASMUS+ Coordinator

Amity University