VEDANT PANCHAL

AI/ML Engineer and Full-Stack Developer

Bangalore, IN.

About

Highly skilled AI/ML Engineer and Full-Stack Developer with proven expertise in building production-ready ML systems, LLM-powered applications, and scalable backend services. Adept at end-to-end model deployment, from real-time predictive maintenance pipelines to advanced RAG systems and multi-agent AI frameworks, leveraging deep learning, NLP, and MLOps to drive impactful solutions.

Work

IIIT Bangalore - CTRI-DG

|

Software Engineer Intern

Bangalore, Karnataka, India

→

Summary

Built and enhanced college administration systems and HRMS, focusing on modular and scalable solutions to improve operational efficiency and traceability.

Highlights

Developed a domain-agnostic Universal HRMS, streamlining the full employee lifecycle including recruitment, onboarding, and leave management, with runtime-configurable approval hierarchies using Spring Boot and Keycloak IAM.

Engineered two new critical modules, Exam Management and Placement Management, for the College Administration System using Spring Boot microservices, integrating via REST APIs with a shared PostgreSQL database, enhancing system traceability from requirements through test cases.

HUDE Labs IT Solutions Pvt. Ltd.

|

Software Engineer Intern

Bangalore, Karnataka, India

→

Summary

Developed and deployed production-level full-stack web applications for scalable e-commerce systems, ensuring robust CI/CD pipelines.

Highlights

Built production-level full-stack web applications using React.js, Node.js (Express.js), and MongoDB for scalable e-commerce systems, designing RESTful APIs for business logic and third-party integrations.

Streamlined multi-service application deployment by containerizing with Docker and Docker Compose, establishing consistent CI/CD pipelines across development and production environments, improving deployment reliability.

Education

CSPIT, Charusat University

Anand, Gujarat, India
→
Bachelor of Technology Artificial Intelligence & Machine Learning Grade: 8.6/10 CGPA

Courses

Advanced Machine Learning

Deep Learning

Neural Networks

NLP

Data Analysis & Visualization

Database Management

Cloud Computing

Skills

Data & Analytics

Power BI, Tableau, Pandas, NumPy, Apache Kafka, MLflow, Airflow.

Languages

Python, JavaScript, Java, C++, C, R, HTML, CSS, SQL.

AI & Machine Learning

PyTorch, Scikit-learn, XGBoost, LightGBM, LangChain, HuggingFace, RAG, NLP, Deep Learning, MLOps.

Frameworks & Libraries

React.js, Next.js, Node.js, Express.js, FastAPI, Spring Boot.

Databases

PostgreSQL, MongoDB, TimescaleDB, Redis, DuckDB, MySQL, SQLite, ChromaDB.

DevOps & Cloud

Docker, Docker Compose, Linux, Nginx, GitHub Actions, AWS, Google Cloud Platform (GCP).

Developer Tools

Git, GitHub, Postman, Keycloak, Groq, REST APIs.

Projects

AI-Powered Predictive Maintenance System

→

Summary

Engineered a real-time predictive maintenance platform monitoring 20 critical oncology equipment units at Zydus Pharma, streaming sensor data via Apache Kafka into TimescaleDB and deploying ensemble ML models (XGBoost, LightGBM, PyTorch DNN) with MLflow experiment tracking. Orchestrated full MLOps infrastructure using Docker Compose, Grafana dashboards, Airflow-based automated retraining DAGs, and MinIO artifact storage to minimize unplanned equipment downtime.

LLM-Powered Call Quality Analyzer

→

Summary

Built an end-to-end NLP and LLM analytics platform ingesting call transcripts, scoring agent performance via LLaMA 3.1 (Groq/LangChain), and delivering insights through role-specific React dashboards with a multi-model pipeline: distilBERT sentiment, KMeans clustering, Random Forest prediction, and Isolation Forest anomaly detection. Implemented a RAG chatbot using ChromaDB and HuggingFace sentence-transformers for natural language queries over transcripts, backed by a Bronze-Gold ETL pipeline in DuckDB.

AI-Powered Cloud Development Environment (Orion IDE)

→

Summary

Built a zero-install browser-based IDE using Google Drive as a cloud filesystem, supporting sandboxed execution across 18 languages via Piston API, backed by a 7-service microservices architecture (API Gateway, Auth, Drive, Editor, Execution, Agent, Notification) with JWT auth and Redis pub/sub for real-time SSE streaming. Engineered a permission-first multi-agent AI pipeline with 6 specialized agents (Planner, Designer, Implementer, Reviewer, File, Run) powered by LLaMA 3.3 70B (Groq) and DeepSeek Coder V2 (OpenRouter).