Hello, I'm

Manav Patel

Passionate about AI & ML

Specializing in Generative AI, Deep Learning, and Computer Vision. Transforming data into intelligent solutions.

View Work Contact Me
Manav Patel

About Me

I am a passionate Data Scientist currently pursuing my Master of Science in Data Science at SVKM’s NMIMS, Mumbai. With a strong foundation in AIML from Gujarat University, I thrive on solving complex problems using cutting-edge AI technologies.

My expertise spans across Generative AI, Computer Vision, NLP, and Predictive Modeling. I have a proven track record of building real-time AI systems, from interview bots to road accident detection pipelines.

2024 - 2026

M.Sc. Data Science

SVKM’s NMIMS, Mumbai

CGPA: 9.0 (FY)
2021 - 2024

B.Sc. AIML

Gujarat University, Ahmedabad

CGPA: 8.5

Technical Arsenal

PythonRSQLPower BI Deep LearningGenerative AINLP Computer VisionOpenCVNeo4j MongoDBGitRAG Hugging FaceDockerAWS

Work Experience

Jun 2025 – Aug 2025

Gen AI Intern

SYBGEN Inc

  • Developed a real-time AI interview system using Whisper, Gemini, Kokoro TTS, and OpenAI models.
  • Simulated voice-based coding and behavioral rounds with dynamic follow-up questions.
  • Generated automated performance reports, saving 10+ hours/week of manual review.

Featured Projects

NETRAI

NETRAI: Road Accident Detection

Dec 2024 – Apr 2025
  • Developed a real-time road accident detection system using YOLOv12 for CCTV footage analysis, achieving 83.3% F1 Score and 0.76 mean IoU.
  • Integrated severity analysis with Gemini 2 Flash and stored annotated accident data in Firebase for immediate access.
  • Deployed end-to-end pipeline using S3 for model storage, Elastic Beanstalk for hosting, and CodePipeline for CI/CD automation.
YOLOv12Gemini 2 FlashFirebaseAWS
Audio Sentiment

Audio Sentiment Analysis & Diarization

Apr 2025 – Jun 2025
  • Built a real-time audio analysis system using Whisper for transcription, Pyannote for speaker diarization, SpeechBrain ECAPA for speaker identification.
  • Fine-tuned DistilBERT for sentence-level sentiment classification, achieving a 10% improvement in accuracy in detecting emotional tone.
  • Mapped sentiments and transcriptions to individual speakers, enabling clear, speaker-specific emotional insights from multi-speaker conversations.
WhisperPyannoteSpeechBrainDistilBERT
Multimodal RAG

Context-Aware Multimodal RAG

Jan 2025 – Mar 2025
  • Built a UPSC-focused RAG system combining OCR and VLM (Qwen2.5-VL) to extract structured insights from multimodal documents using Docling.
  • Integrated Milvus for semantic retrieval within a Docker-containerized RAG pipeline, enabling linear horizontal scalability and sub-5ms query latency.
  • Designed a smart context management system with LangGraph and tracked student progress and concept mastery in Neo4j.
Qwen2.5-VLDoclingMilvusLangGraphNeo4j
Currency Forecasting

Currency Exchange Forecasting

Jan 2025 – Mar 2025
  • Forecasted EUR/INR exchange rates with 3.59% MAPE using Prophet and modeled volatility with GARCH/EGARCH (best AIC: 921.1).
  • Benchmarked ARIMA/SARIMA as baseline forecasts and applied Vector Autoregression (VAR) to capture inter-currency dependencies.
ProphetGARCHARIMAVAR
Credit Risk

Credit Risk & Loan Analytics

Nov 2024 – Dec 2024
  • Built an end-to-end ETL and analytics pipeline in Python and Power BI to visualize loan default data and support strategic risk assessment.
  • Identified key default drivers using Chi-Square and Kruskal–Wallis tests, strengthening credit risk models and lending policies.
  • Designed a risk categorization framework in Python and DAX to calculate LGD and deliver actionable KPIs for portfolio monitoring.
PythonPower BIETLDAXStatistics

Awards & Certifications

Kaggle Notebook Expert

Recognized for high-quality data science notebooks.

1st Place - Data Insight

Gujarat University Competition Winner.

Best Innovative AI Solution

VenusMoon Education Challenge 2024.

Data Science Masters

PW Skills (Jan 2024)

Get In Touch

Email

manavp571@gmail.com

Phone

+91 8530893089

Location

Mumbai, India