Data Scientist

Vindhya Regonda

+91-996******* ***********@*****.*** Hyderabad, India

Data Scientist with 2+ years of experience in Computer Vision, ML/DL and GenAI specializing in developing advanced end-to-end production-ready AI solutions and scaling them into business-ready products.

Work Experience

SensEarlyAI

Data Scientist Sep 2024 – Present

• Developed a clinical document analyzer using an agentic AI pipeline to extract and organize medical documents into a structured timeline, enabling doctors to visualize patient history at a glance. Engineered a liver-transplant-specific scheduler that automatically generates a personalized consultation/lab/scan timeline from the transplant date and auto-fills it from uploaded lab report PDFs with 94% accuracy. Incorporated a human-in-the-loop review layer that reduced manual review time by 70%, flagging edge cases — unit, keyword, LOINC code, and reference range mismatches — for doctor confirmation before values are accepted.

• Enabled scalable smartphone diagnostics for 100+ users to predict jaundice based on sclera analysis from user-uploaded images having more than 95% accuracy. Engineered a multi-stage image-quality and anatomical validation layer that rejects blurred/low-quality images, non-human faces, closed eyes, spectacles/occlusions, unsuitable framing, poor lighting/color conditions, and images with insufficient sclera visibility, ensuring adequate scleral pixels contribute to jaundice prediction; incorporated robust preprocessing, object detection, segmentation gates, and feedback loops to improve valid image submissions by 40% under real-world capture conditions.

• Reduced clinical review time by 15+ hours/week. Built a scalable clinical intelligence pipeline that transforms raw patient vitals into explainable, prioritized, actionable insights. By combining statistical modeling, trend analysis, and contextual rule evaluation, the system reduces alert noise, highlights meaningful health changes, and enables proactive care escalation in remote patient monitoring.

• Achieved 96% accuracy in clinical numerical data extraction by building a multi-stage pipeline processing 2000+ medical device images (BP, thermometer, weighing scale, pulse oximeter), combining SIFT-homography correction, screen/digit segmentation, connected-component digit extraction, and ResNet-18 classification to handle perspective, rotation, lighting, and screen noise.

• Personalized Diet Plan Generation & Adaptive Nutrition Engine. Designed a multi-stage AI pipeline to generate personalized diet plans using clinical condition, target protein/calorie intake, daily routines, dietary preferences, cuisine, eating barriers, special dietary requirements and patient feedback. Built iterative feedback loops that evaluate patient compliance and dynamically update subsequent diet plans, enabling continuous personalization based on adherence and changing nutritional requirements.

• Context-Aware Chatbots (AskOrca & Nutribot) Developed patient-context-aware AI chatbots for remote patient monitoring and nutrition support, leveraging patient history, clinical data, dietary preferences, and health context to provide personalized responses, recommendations, and actionable guidance. Implemented symptom-severity detection, escalation workflows, and safety guardrails to identify clinically important symptoms, notify the appropriate care team, and prevent unsafe or unsupervised medical recommendations.

• Patient Compliance & Remote Health Monitoring Pipeline. Developed patient compliance pipeline that dynamically tracks daily health tasks including BP, temperature, weight, smartwatch synchronization, questionnaires, symptom updates, and image submissions. Automated collection and validation of multimodal patient data for continuous clinical monitoring, integrating wearable, questionnaire, vital-sign, and computer-vision signals to support early identification of health risks and improve adherence to remote-care protocols.

• Architected scalable data pipelines on GCP, saving 10+ hours/week of manual effort by automating third-party data ingestion and ensuring 99.9% data integrity via a ’supercron’ job that back-calculated anomalies from late patient device syncs.

Teaching Assistantship at IIIT-Delhi New Delhi, India Machine Learning Sep 2022 – May 2024

• Conducted tutorials and evaluated student assignments, exams, quizzes, and projects for the courses - ML/DL Education

Indraprastha Institute Of Information Technology, Delhi New Delhi, India M.Tech. Computational Sciences; [Transcript] CGPA: 9.23/10.00 Aug 2022 – Jun 2024 Bombay College of Pharmacy Mumbai, India

B.Pharm.; [Transcript] CGPA: 8.81/10.00 Aug 2018 – May 2022 Skills

Languages: Python, SQL, R

Computer Vision: OpenCV, YOLO, U-Net, Image Segmentation, Object Detection, OCR, Image Processing, SIFT, Homography, NCC, FLANN, CLAHE, RANSAC, exposure correction, seam finding, blending, panorama ML/AI Frameworks & Libraries: PyTorch, TensorFlow, Scikit-Learn, Hugging Face, MLflow, LangChain, NumPy, Pandas, Predictive Modeling, Classification, Regression, Feature Engineering, Model Evaluation GenAI & LLMs: LLMs, Generative AI, Agentic AI, RAG, Prompt Engineering, Claude, Gemini, OpenAI, NLTK, Spacy

Cloud & Data Engineering: Google Cloud Platform (GCP), Firestore, Data Pipelines, SQL, API Integration Developer Tools: Git, GitHub, Streamlit, Jupyter Notebook, Google Colab, VS Code, PyCharm Tools: GitHub, Google Cloud platform, Jupyter Notebook, Google Colab, VSCode, PyCharm, Google Workspace Tools Publications

Overlapping gut microbiome signatures in aging and disease are characterized by enrichment of medication-associated oral microbes in the gut (WILEY) Nov’22 – May’24

• Tech Stack: R, Python, Linux, Statistical Analysis, Data Analysis and Visualization, Hypothesis Testing

• Comprehensive characterization of non-antibiotic medication associated gut microbiome modules at functional and mechanistic levels. [Report]

Projects

Retail Shelf Panorama Stitching using Computer Vision GitHub

• Tech Stack: Python, OpenCV, SIFT/RootSIFT, RANSAC, Homography, SciPy, NumPy, Streamlit

• Developed a robust CV pipeline using SIFT/RootSIFT, RANSAC, and homography optimization to stitch unordered retail shelf images into a full-resolution panoramas. Achieved 1.42 px reprojection RMS across a 5-store benchmark, addressing repetitive product patterns, incorrect feature matches, uneven exposure, and shelf distortions to improve alignment reliability and image completeness. LLM-RAG Streamlit News Insights GitHub WebApp

• Tech Stack: Langchain, Huggingface, FAISS, Gemini-Pro, Streamlit, unstructured

• Generative-AI was employed to extract insights from news content (links) and was implemented through a Streamlit interface on Google Colab for real-time application. Automated Classification of Product Documents into Industry-Specific Categories GitHub WebApp

• Tech Stack: OCR tools, Tesseract, PDFMiner, XGBoost, DL, sentence-transformer, Huggingface, Streamlit

• Developed an end-to-end pipeline to extract text from PDF urls and constructed an ML model to classify them using extracted features, achieving more than 95% accuracy. Deployed an inference pipeline and hosted application for real-time classification.

MRI-Scan Image Segmentation using YOLOv10 GitHub

• Tech Stack: Python, YOLO, TensorFlow, Roboflow

• Fine-tuned YOLOv10 model for accurate segmentation of brain tumors and eyes in MRI scans, enabling early detection and precise annotation of potential tumor regions with greater than 90% precision. Social Implications of NLP-Driven Healthcare Technologies: Personalized Medicine GitHub

• Tech Stack: PyTorch, BERT, Clinical-BERT, Web scraping - BeautifulSoup, fine-tuning, pre-training

• Developed a BERT model and fine-tuned on the training data that outperformed the existing models

(Clinical-BERT) for clinical purposes, enabling doctors to diagnose diseases and prescribe tailored medications based on patient metadata and drug reviews.

Awards & Achievements

• Awarded the Commendable Green Chemistry Award by the American Chemical Society in recognition of impactful green chemistry activities.

• Awarded a Contributor Certificate by IISER-Pune, NCCS, and Persistent Systems for accurate annotation and curation of scientific research papers on the MANAV Platform in compliance with prescribed standards.

• Secured 2nd place in National Level Poster Competition organized by IES’s Management College & Research Center. Extra-Curriculars / Leadership

• Member of Design team, Convocation Committee 2023 - IIIT Delhi (Jun 2023 - Mar 2024)

• Teacher volunteer at U&I Trust (Jul 2023 - Mar 2024)

• Mentor of newly admitted M.Tech. Students. (Aug 2023 - May 2024)

• Class Representative of U.G., B.Pharm., Mumbai University (Jul 2021 - May 2022)

• Co-founder & Editorial Head of ACS-BCP SC (Student Chapter) (Jul 2020 - Jul 2021)

Contact this candidate