Technical Skills
Languages
JavaScript
TypeScript
Python
C++
Backend & APIs
Node.js
NestJS
Express.js
GraphQL
Frontend
React
Redux
HTML5
CSS3
Databases
MongoDB
SQL Server
Cloud & DevOps
AWS
Docker
AI & ML
TensorFlow
PyTorch
RAG
LLMs
Tools
Postman
Swagger
Apollo Studio
✦ my journey
Work Experience
Software Engineer (Full-stack)
EB Pearls Pvt. Ltd.
Nov 2022 – Jul 2024 Sydney, Australia (remote)
- ▸Developed and delivered production-ready web applications, making architectural decisions on data modeling and API structure by collaborating with cross-functional teams.
- ▸Designed and implemented robust RESTful and GraphQL APIs with React and Redux frontend.
- ▸Improved application performance using database indexing, aggregation pipelines, and query optimizations.
- ▸Implemented role-based access control (RBAC) and authentication workflows to secure access division across various user levels.
- ▸Integrated Twilio (OTP/SMS), SendGrid, AWS SES for email, and AWS S3 for storage.
- ▸Enhanced code maintainability using TypeScript with centralized error handling.
Software Engineer Trainee
EB Pearls Pvt. Ltd.
Apr 2022 – Oct 2022 Sydney, Australia (remote)
- ▸Implemented node-cron scheduled jobs and Mongoose geo-spatial queries in a SAAS application.
- ▸Built responsive web applications and admin dashboards with MERN stack and NestJS with GraphQL.
- ▸Utilized Postman, Swagger, and Apollo Studio for API development and testing.
Publication
Uncertainty-Aware LLM-Guided Policy Shaping for Sparse-Reward Reinforcement Learning
✦ Accepted — IEEE CAI 2026Proposed a calibrated reinforcement learning framework using a fine-tuned BERT model, Proximal Policy Optimization, Monte Carlo dropout, and A* based oracle in a custom UnlockPickup MiniGrid environment. Achieved 99.20% success rate, 2009.96 reward AUC, and reduced the average steps to goal from 35.52 to 8.56, outperforming baseline models and improving decision-making in sparse-reward environments.
View PaperLeadership Experience
Entrepreneurial Lead
NSF I-Corps Hub Great Plains
Jan 2025 - Feb 2025 & Jun 2025 – Jul 2025 North Dakota, USA (remote)
- ▸Led 3-member teams in two cohorts to explore AI-driven tools in music and legal domains.
- ▸Conducted 40+ customer interviews applying lean startup and customer segmentation techniques.
- ▸Collaborated with technical and business mentors on ecosystem modeling and value propositions.
✦ things I've built
Featured Projects
AI, Machine Learning & Full-Stack Development
Hackathon, 2nd Place - USD Ignite
PlanMyPlate — AI Meal Planning App
Apr 2026
Developed a multi-agent AI meal planning system powered by Anthropic’s Claude API, delivering personalized weekly meal plans, macro-optimized grocery lists, and cooking schedules tailored to user body composition, dietary preferences, and budget. Built and deployed a full-stack solution with a FastAPI backend and React/TypeScript frontend within a 48-hour sprint.
PythonReactFastAPITypeScriptClaude API
Machine Learning & AI
RAG-Based E-Commerce Price Prediction System
Aug 2025 – Dec 2025
Developed a prediction system to estimate expected customer ratings (regression) and launch risk level (classification) for Amazon datasets. Compared Random Forest, Linear/Logistic Regression, XGBoost, and LightGBM. Used SHAP to interpret how different factors impact a product's success.
Machine LearningExplainable AI (SHAP)Predictive AnalysisFeature Engineering
Information Storage & Retrieval
Protein Chatbot with FAISS & LLM
Aug 2025 – Dec 2025
RAG-based web application with Sentence Transformer and FAISS for semantic search, powered by Llama3.2 for accurate protein information.
RAGNLPFAISSPythonDocker
Distributed System
Federated Learning in Distributed Systems
Jan 2025 - May 2025
Performed a comparative analysis of federated learning algorithms (FedAvg, FedProx, Cyclic Weight Transfer, Differential Privacy) in a distributed systems environment using the MNIST dataset.
Federated LearningPythonPyTorch
Computer Vision
Comparative Analysis of Object Detection Models in Autonomous Driving
Sept 2024 - Dec 2024
Pre-processed the KITTI dataset and analyzed the trade-offs between YOLOv5 and Faster RCNN for speed vs accuracy. Optimized the loss function with classification loss, bounding box regression, and objectness loss.
Computer VisionDeep LearningPyTorchTensorFlowPython
Machine Learning
Dental Caries Segmentation for Panoramic X-ray Images
Sept 2024 - Dec 2024
CNN-based model for automated caries detection in panoramic X-rays, creating clinical decision support tools. Segmented dental caries using U-Net architecture with ResNet34 encoder for 233 panoramic X-ray images.
Medical Image AnalysisComputer VisionTensorFlowPythonDeep Learning