About Me
I’m a Research Scientist (AI/ML) at Drevol with a Bachelor's in Computer Science from UC Davis with deep expertise in building scalable, cloud-native AI systems. As a AI Research Engineer, I’ve architected containerized, load-balanced pipelines using RAG-based LLMs on Azure and AWS. In my UC Davis research role, I led TinyML and Health-LLM initiatives—delivering a quantized ECG anomaly detector (IEEE ISCAS accepted) and a personalized Health LLM via fault-tolerant, event-driven data pipelines.
Key Expertise & Technologies
Cloud & Infrastructure
- AWS (EC2, S3, Lambda, ECS/EKS)
- Azure (VMs, Functions, AKS)
- GCP (Compute Engine, Cloud Functions)
- Terraform, CloudFormation
- Docker, Kubernetes, Serverless Architectures
- Load Balancing, CDN
AI/ML & Data Engineering
- Core ML models, RAG LLMs, TinyML (edge-computing)
- PyTorch, TensorFlow, XGBoost, SHAP, LIME
- Apache NiFi, FHIR, Kafka
- GridSearchCV, Multilabel Classification
- Event-Driven Pipelines
Backend & Microservices
- C#, .NET Core, Python (Flask, Django)
- Node.js, SQL (PostgreSQL, SQL Server)
- NoSQL (MongoDB), Prisma
- Microservices, RESTful & gRPC APIs
- CI/CD (GitHub Actions), Containerized Workflows
Frontend & Web
- Next.js, React, Tailwind CSS
- HTML5/CSS3, JavaScript/TypeScript
- Responsive UI, Vercel, Render
Developer Tools & Collaboration
- Git/GitHub, Docker Compose, JIRA
- VS Code, PyCharm, Eclipse, Jupyter
- CI/CD Pipelines, Agile/Scrum Practices
I’m passionate about microservices, serverless architectures, and advanced data engineering to drive measurable impact, and about driving AI-powered innovation in cloud environments—building robust ML pipelines, automating workflows, and deploying high-availability services. Let’s connect!
Experience
Aug 2024 — Present Drafted the research agenda on LLM/RAG and multimodal models in PyTorch—establishing internal benchmarks
Defined success metrics and an MVP roadmap for AI agents; set up prompt/agent CI/CD and regression evals to enable rapid, measurable releases
Partnered with engineering to set up agent orchestration (LangGraph) and LLM observability
- Machine Learning
- Data Science
- Software Engineering
- Full-Stack Development
Apr 2025 — Jun 2025 Designed and implemented privacy-preserving, distributed log-analysis workflows using retrieval-augmented \ (RAG) LLMs , cutting error-triage time by 65% and boosting resolution accuracy by 40%
Designed and optimized scalable, load-balanced AI-driven test orchestration pipelines leveraging Kubernetes, \ Docker, and Azure cloud services, increasing throughput by 50% and accelerating release cycles by 25%
Engineered and refined modular automation solutions in C#, .NET, and SQL, leveraging containerized, \ microservice-friendly architecture to increase task efficiency by 20%
Developed and deployed containerized AI solutions using image recognition and NLP (Natural Language \ Processing), reducing manual effort by 25% while ensuring high availability and horizontal scaling
- Machine Learning
- Data Science
- Software Engineering
- Full-Stack Development
Feb 2024 — Present Engineered energy-efficient ECG classification using a quantized TinyML Random Forest (92.8% accuracy) and an event-driven architecture with adaptive burst-mode data collection, extending wearable battery life from 14 days to over a month.
Enhanced anomaly detection to 93.6% accuracy via advanced feature engineering and strategic hyperparameter tuning with GridSearchCV on resource-constrained devices.
Optimized deep learning for time-series data using SHAP and LIME, achieving a 7% accuracy boost and enabling hybrid offloading of complex multilabel classification to server-side CNN and XGBoost models for comprehensive health monitoring.
- Machine Learning
- Deep Learning
Oct — Dec 2021 Executed comprehensive data preprocessing and transformation on inconsistent datasets, employed classNameification and regression techniques for training supervised learning models to achieve an impressive 98% accuracy.
Leveraged Matplotlib and Seaborn for advanced data visualization, facilitating data-driven decision-making.
- Machine Learning
- Python
Jan — Apr 2024 Led the Summarizit project team to develop an innovative web application that analyzes and summarizes extensive video content, resulting in a 40% increase in user productivity and information accessibility.
Leveraged NextJS (Typescript) for the frontend, Express (Typescript) for the backend, and MongoDB for database management, implementing CRUD routes to ensure robust, scalable, and efficient data management.
- Typescript
- NodeJs
- MongoDB
- React
- ExpressJs
- NextJS
Sep 2023 — Present Led AI/ML project initiatives, including the development of an advanced music recommendation system that improved user engagement by 35%, a medical chatbot for injury diagnosis that achieved 90% accuracy in providing preliminary assessments, and a strategic poker bot that outperformed human players in 70% of simulated games.
- Python
- Javascript
- NodeJs
- MongoDB
- React
- ExpressJs
- NextJS
Projects
CorpCred
CorpCred is an AI-powered Corporate Credit Rating Calculator that predicts a company's credit rating based on its financial ratios. This project applies machine learning to financial analysis, helping businesses, investors, and analysts make informed credit decisions.
- Neural Networks
- Next.js
- Vercel
- Render
- TailwindCSS
EduParse
Created a chrome extension which provides users with practice exams with a variety of questiosn including short and long answer questions, multiple-choice questions, nuemricals, etc. based on provided lecture notes from users.
- Python
- Django
- ChatGPT API
Scream Detection Deep Neural Network
The Scream Detection AI/ML Model is an innovative system developed to enhance the safety and security of our college community. The project aims to detect distress calls and screams in real-time, enabling immediate response and providing assistance to individuals in danger.
- Python
- TensorFlow
- SciPy
- Sklearn
Portfolio Website
Personal Website for showcasing my experience, projects and my jounrey as a software developer.
- TailwindCSS
- Javascript
- Vercel
Breast Cancer Detection
Developed Breast Cancer Detection machine learning models leveraging six different ML models, including Logistic Regression and Decision Tree classNameifier to accurately classNameify breast cancer as benign or malignant with an average accuracy rate of 98%.
- Gatsby
- Styled Components
- Netlify