Purva Soni

AI Researcher

India2 yrs 2 mos experience

AI EnabledAI ML Practitioner

Key Highlights

Reduced diagnostic report turnaround time to under 4 minutes.
Achieved 97% accuracy in anemia detection.
Delivered production AI systems in healthcare and business.

Stackforce AI infers this person is a Healthcare-focused AI Engineer with expertise in Machine Learning and Data Science.

Contact

Skills

Core Skills

Machine LearningDeep LearningData AnalysisBusiness AnalysisResearch Skills

Other Skills

Data SciencePredictive ModelingStatistical & Probabilistic ModelingOptimizationAlgorithm DevelopmentAdvanced AnalyticsQuantitative ResearchPython (Programming Language)Statistical ModelingProblem SolvingGenerative AIArtificial Intelligence (AI)Applied ResearchAmazon Web Services (AWS)Google Cloud Platform (GCP)

About

I help teams turn messy, high-dimensional data into reliable decisions and production-ready models. With a background in Machine Learning, Data Science, and Applied Statistics, I specialise in solving problems where accuracy, scalability, and real-world constraints matter—whether that’s predictive modelling, optimisation, or probabilistic decision-making. What I bring to your team: • Predictive, statistical, and probabilistic models that reduce uncertainty and improve business or clinical outcomes • End-to-end ML systems—from data pipelines and feature engineering to deployment, monitoring, and optimization • Advanced analytics and algorithm development grounded in applied mathematics, statistics, and quantitative research I’ve delivered production AI systems in healthcare and business environments—cutting turnaround times, improving signal-to-noise in data, and deploying models that stakeholders actually trust and use. I help organisations turn complex challenges into scalable, real-world impact 🎯✨.

Experience

2 yrs 2 mos

Total Experience

8 mos

Average Tenure

Current Experience

Sigtuple

AI-Engineer 1

Jul 2024 – Feb 2026 · 1 yr 7 mos

Built end-to-end Machine Learning and Computer Vision pipelines for AI-embedded in-vitro diagnostic devices, automating manual microscopy through robotics, image processing, and deep learning.
Designed and deployed predictive, statistical, and probabilistic models for cell detection, classification, quality assessment, and diagnostic decision support.
Owned the full data science lifecycle—data curation, feature engineering, model training, optimization, validation, deployment, and monitoring in production environments.
Notable Achievements:
Reduced diagnostic report turnaround time to under 4 minutes by developing optimized deep learning pipelines for WBC, RBC, and platelet analysis.
Achieved 97% accuracy in anemia detection by combining statistical modeling, threshold optimization, and predictive analytics on 10K+ annotated cell images.
Deployed production-grade AI systems adopted in nationwide public health programs by implementing model optimization, reproducibility testing, and scalable deployment workflows.
Skills Developed:
Machine Learning, Deep Learning, Data Science, Predictive Modeling, Statistical & Probabilistic Modeling, Optimization, Algorithm Development, Advanced Analytics, Quantitative Research

Machine LearningDeep LearningData SciencePredictive ModelingStatistical & Probabilistic ModelingOptimization+4

Hero dealership

Data Scientist

Jan 2024 – May 2024 · 4 mos

Implemented ML-driven market segmentation to enhance customer targeting and improve sales growth.
Developed predictive demand forecasting models that optimized inventory management.
Strengthened retail analytics to boost profitability and elevate customer satisfaction.

Data AnalysisBusiness Analysis

Indian institute of technology, kanpur

Researcher

Jun 2021 – Sep 2021 · 3 mos · India

Worked on a research paper for verifying an "Analogous operation in elliptic curve cryptography that generated a signature to be used as a range proof of k-times anonymous authentication".
Extracted crucial information from the research paper to define a signature proof of knowledge (SPK) in ECC, to check the stipulated usage limit for a scheme proposed in the paper, demonstrating proactive problem-solving abilities, the SPK5a implementation in C++ was executed to validate its accuracy.
Improved process efficiency to 85% by remoulding the SPK to validate its correctness in Elliptic Curve Cryptography.
Concluded RSA signatures are 10 times slower than ECC for private key operations. ECC encryption with significantly lower key sizes can provide equivalent cryptographic strength related to an RSA based system’s

Problem SolvingResearch Skills