Experience
Applied AI impact
- Fine-tuned Meta SAM2 for segmentation on medical imaging data to support automated segmentation and annotation workflows.
- Developed preprocessing pipelines for medical imaging data quality and ML training efficiency.
- Analyzed transportation, OEM, and last-mile challenges to surface technical priorities.
- Translated stakeholder needs into technical requirements for product development.
- Built KPI dashboards to provide real-time operational decision support.
Reference letter
Education
Academic foundation
- GPA: 2.3 | Focus: Data and Intelligence
- Design and evaluation of a neurosymbolic child-safe AI architecture for ages 4-7.
- Hybrid neural semantic parser with symbolic safety governor for traceable control.
- Benchmarking against NVIDIA NeMo Guardrails with age-specific interaction datasets.
- Automated SysML 2.0 diagram generation and explanation using GPT-4 Vision.
- Streamlit interface to streamline system modeling for engineering teams.
Project video
Research
Research-driven work
SAM2 Medical Imaging Segmentation
Feb 2024 - Present
Fine-tuned Meta SAM2 for segmentation on medical imaging data to support automated annotation workflows in minimally invasive surgery.
Neurosymbolic Child-Safe AI Architecture
May 2026 - Oct 2026
Designing a hybrid neurosymbolic architecture combining neural semantic parsing with deterministic
symbolic safety controls for children aged 4-7.
GenSys: SysML 2.0 Automation
Oct 2023 - Aug 2024
Automated SysML 2.0 diagram generation, modification, and explanation from prompts and images.
View demo
Open Source
Community contributions
DataFog - PR #148
Contributed German-specific PII detection and locale-gated regex behavior to an open-source Python PII detection and redaction library, enabling opt-in support via locales=["de"].
View PR
Passion Projects
Independent explorations
VisionExpert-RAG
Visual RAG system using ColPali and Mistral-7B for technical document auditing.
GitHub
ACC-RAG Architecture
Developed an Adaptive Contextual-Retriever for local systems, fusing text with OCR diagram labels to extract all details from engineering manuals.
GitHub
Multimodal Fracture Detection
Built an end-to-end fracture screening prototype that combined CNN-based X-ray detection with MedGemma explanations.
GitHub
Brain Tumor Detection with YOLO
YOLO-based localization model using PyTorch and OpenCV.
GitHub
Handwritten Digit Recognizer
Automated CI/CD pipeline that trains and deploys MNIST digit classification models.
GitHub
Skills
Core strengths
Domains
Machine Learning
GenAI
Agentic AI
Computer Vision
Neurosymbolic AI
Languages
Python
R
Java
SQL
LaTeX
AI/ML
PyTorch
LangChain
NeMo Guardrails
YOLO
CNN
RAG
SysML
Tools
Docker
Git
FastAPI
Streamlit
Power BI
OpenCV
Pandas
CI/CD
Languages Spoken
English (Fluent)
German (A2, Learning)
Hindi / Gujarati (Native)
Certificates
Continuous learning
Volunteering
Community involvement
Interests
Beyond the desk
Community Building
Public Speaking
Table Tennis
Volleyball
Cycling
