Core ML, deep learning, LLM/RAG systems, model evaluation, and deployment workflows.
Software Engineer • AI/ML • Data Engineering
Hi, I’m Pathik Viramgama.
I build full-stack products, data platforms, and AI-powered systems that are fast, reliable, and useful in real-world environments.

Engineer with a product mindset.
USC MSCS graduate building across backend, cloud, frontend, AI/ML, and modern data workflows.
About
Engineer who enjoys turning messy ideas into working systems⚙⚙.
I’m Pathik — a software engineer who builds practical products across full-stack development Full-Stack Development React Next.js Node.js Django FastAPI HTML CSS TypeScript , AI/ML systems AI / ML Systems OpenAI PyTorch TensorFlow Keras Scikit-learn Hugging Face BERT OpenCV , data engineering Data Engineering Snowflake Databricks Azure SQL SSMS ETL Pipelines Power BI Power Apps , and cloud infrastructure Cloud Infrastructure AWS Azure GCP Docker Kubernetes CI/CD Firebase Git .
I enjoy taking ideas from rough concept to polished product — designing the architecture, writing clean code, testing deeply, and shipping something people can actually use.
Reliable pipelines, lakehouse layers, analytics tables, and dashboards.
Clean interfaces, scalable APIs, authentication, and cloud deployment.
Skills
Technical Toolkit
Tools I use to build, scale, automate, and ship production-ready systems.
Languages
Think → Type → Compile → Ship
Frontend & Product
Design → Component → Motion → Experience
Backend
Request → Validate → Query → Respond
AI / ML
Prompt → Retrieve → Rerank → Generate
Cloud & DevOps
Code → Container → CI/CD → Cloud
Data Engineering
Ingest → Transform → Warehouse → Dashboard
Experience
Recent Work
Database Engineer · California Creative Solutions↗
Built end-to-end data engineering pipelines using Snowflake, Databricks, Azure, and SQL. Automated ingestion, transformation, and reporting workflows, improving processing efficiency by 35%.
- Optimized SQL, stored procedures, SSMS workflows, and Snowflake ETL logic.
- Reduced manual data preparation time by 40% while improving reporting accuracy.
- Built Power BI and Power Apps solutions for real-time visibility and self-service reporting.
Software Developer · VibeSea↗
Designed backend architecture, database schemas, and production APIs for a fast-moving startup building anti-cheat and proctoring workflows.
- Built scalable detection systems using Node.js, Django, DynamoDB, PostgreSQL, and Amazon S3.
- Authored OpenAPI 3.0 REST APIs and optimized services for maintainability and auditability.
- Deployed a Next.js website through Vercel CI/CD, improving load times by 20%.
Software Developer Intern · WorkUp↗
Improved Kotlin mobile app reliability and backend integration while working closely with a 7-member team and company leadership.
- Reduced crash rates by 30% through Kotlin, JetBrains, and Retrofit optimizations.
- Implemented caching using Caffeine and MutableMaps, reducing GCP costs by 8%.
- Supported production environments with 99.99% uptime across GCP servers and Docker containers.
Software Developer (ML) · Orail Services↗
Developed AI-driven MERN applications with machine learning integrations, computer vision workflows, and FastAPI-backed reverse image search systems.
- Built AI-driven MERN apps and REST APIs, increasing engagement by 20% and conversion by 37%.
- Integrated OpenCV/deep learning face recognition, reducing login time by 60%.
- Built FastAPI reverse image search with ResNet101, KNN, and cosine similarity, achieving 94% accuracy.
ML Intern · Meteorological Department↗
Built climate forecasting models and real-time inference pipelines for large-scale satellite data analysis in collaboration with ISRO.
- Built LSTM/BiLSTM models with 92.6% accuracy, improving climate forecasting efficiency by 18%.
- Processed 16TB+ IMD-A/IMD-C satellite data with ISRO-integrated models, reducing inference latency by 40%.
- Supported beta deployment across North India with 87% real-world forecasting accuracy.
Projects
Selected Projects
Grouped by focus area, with each project card linked for GitHub, live demos, or case studies.
AI / ML
AutoApply AI Agent System
Designed an 11-agent CrewAI workflow powered by LLM Router BYOK, LiteLLM, and Tavily, enabling automated job description parsing, company research, resume-job match scoring, strategy generation, resume tailoring, truthfulness review, recruiter outreach, and interview preparation with real-time agent progress streaming.
View project → AI / ML Currently Active Last updated: loading...Machine Learning Algorithms from Scratch
Implemented machine-learning and deep-learning algorithms from scratch using only pandas and NumPy, with reusable code examples and experiment notebooks.
View project → Multi-Agent AI Under Devlopment Last updated: loading...FinanceDoc AI
Built a multi-agent financial document analyzer for invoice and purchase order reconciliation using PDF extraction, Pydantic validation, LangGraph workflow orchestration, risk scoring, and human-in-the-loop review.
View project → Multi-Agent AI Under Devlopment Last updated: loading...HealthAgent
Built a synthetic healthcare multi-agent AI system for medication adherence monitoring with ML-based risk prediction, patient segmentation, refill checks, intervention recommendations, escalation review, and audit logging.
View project → AI / MLVideo Sentiment Analysis Model
Built the multimodal sentiment model pipeline combining video, audio, text, Whisper transcription, BERT-based NLP, and SageMaker-backed model workflows.
View project → AI / MLOptigo Apps ML Platform
Dockerized ML models on GCP with real-time REST APIs, customer value scoring, and high-uptime MERN integration.
View project → AI / MLCustomer Rating ML System
Developed a machine-learning prototype to predict customer quality scores from order activity, category spend, pricing, labor charges, and commercial metrics for prioritization decisions.
View project → AI / MLReverse Image Search
Engineered a reverse image search system for a custom database using TensorFlow ResNet50 and OpenAI CLIP, exposed through FastAPI and a React frontend on Linux-based AWS instances.
View project → AI / MLFuture Air Pollution Prediction
Studied historical air-pollution patterns and advanced the work toward predicting future pollution levels using XGBoost and Principal Component Analysis.
View project → AI / MLFake News Classification Comparative Study
Applied deep learning and machine-learning algorithms to compare fake-news classification behavior across LSTM, BiLSTM, BERT, GPT-2, RoBERTa, and classical ML baselines.
View project → AI / MLFake News Classification Novel Approach
Researched grammatical patterns for fake-news detection and developed a POS-tag occurrence approach to create decision-tree features, reaching 89% accuracy in raw format.
View project →Web & Backend
AutoApply AI SaaS Platform
Built a full-stack SaaS-style job application platform using React, Vite, FastAPI, Neon PostgreSQL, Stripe, JWT auth, and Google OAuth, supporting user accounts, credit-based usage, resume/JD uploads, payment-based credit purchases, saved job history, live progress UI, and DOCX/PDF/Markdown exports.
View project → Web Currently Deployed Last updated: loading...LLM Router
Built an OpenAI-compatible LLM router that lets users connect multiple provider API keys, route chat requests through one local endpoint, automatically fail over across providers, and monitor usage with analytics, request logs, provider health, and rate-limit awareness.
View project → WebVibeCheck Anti-Cheat Interview Backend
Developed backend APIs for a remote interview integrity platform with secure onboarding, scheduled interviews, media uploads, AI vision analysis, alerts, review flows, and Stripe credits.
View project → WebVideo Sentiment Analysis Website
Built the secure SaaS web experience for uploading media, reviewing AI sentiment results, and presenting model outputs through a polished Next.js interface.
View project → WebStock Trading Platform Website
Created a stock-trading web platform using React, MongoDB Atlas, Node.js, and Express, enabling users to create profiles and study trading patterns without financial risk.
View project →App Development
Stock Trading Platform Android Application
Created a detailed stock-trading Android platform using Kotlin and XML layouts with a MongoDB Atlas, Node.js, and Express backend for user profiles and risk-free trading-pattern analysis.
View project → AppDVD Rental System and Business Encircling It
Created a comparative analysis of an old DVD rental system and modern subscription-based technology, combining business analysis with system-design thinking.
View project → AppDino Run Game in Flutter and Flame Engine
Built a graphical Chrome Dino Run-inspired game in a Mario Run format using Flutter, Dart, and the Flame game engine.
View project →Data Engineering
Azure Healthcare Lakehouse Pipeline
Designed a healthcare revenue-cycle data pipeline that ingests SQL, files, and public APIs, applies Medallion architecture, validates data quality, and builds Gold fact/dimension tables.
View project → DataNetflix Azure Lakehouse Project
Built an end-to-end Azure data engineering pipeline for Netflix-style analytics using ADF, Databricks, Delta Live Tables, Unity Catalog, and Power BI-ready Gold tables.
View project → DataReal-Time Azure Streaming Pipeline
Created a live weather streaming pipeline with dual ingestion through Databricks and Azure Functions, Event Hubs, Fabric Eventstream, Eventhouse/KQL, Power BI, and alerting.
View project → DataMDOTSA Azure Data Pipeline
Implemented a metadata-driven transit data pipeline ingesting REST APIs, CSV drops, and on-prem SQL Server through SHIR into ADLS Gen2 and Azure SQL Bronze/Bronze_Hist layers.
View project → DataAzure Databricks Incremental Lakehouse
Built a Databricks Lakehouse pipeline for batch, incremental, and streaming data with Auto Loader, Structured Streaming, Bronze/Silver/Gold Delta tables, and workflow orchestration.
View project → DataWeb Scraping Pipeline
Scalable Python scraping system extracting data from 100+ domains and automating cleaning workflows to reduce processing time by 40%.
View project →Analytics & BI
India Election Results Power BI Dashboard
Designed an interactive Power BI dashboard for election analytics covering party performance, alliances, state-level results, vote share, seat distribution, maps, drill-throughs, and KPIs.
View project → BIHospital Analytics Power BI Dashboard
Built an end-to-end analytics dashboard with data cleaning, modeling, DAX measures, KPI cards, slicers, filters, and polished report design for healthcare/business reporting workflows.
View project →Education
Academic background
A computer science foundation shaped by product engineering, AI systems, and data-focused problem solving.
Master’s
University of Southern California
Master’s in Computer Science · GPA 3.7/4.0
B.Tech
Pandit Deendayal Energy University
B.Tech in Computer Science & Engineering · GPA 3.85/4.0
Recognition
Recommendations & letters of appreciation.
Feedback and appreciation from people I’ve worked with across engineering, academic, and project environments.
“Recognized by Orail Services for strong problem-solving, AI/ML implementation, and practical engineering ownership across Optigo Apps and EVO App. Pathik delivered machine-learning features despite data limitations, including a KNN-based jewellery recommendation system that performed strongly in alpha testing.”
“Recommended for his academic dedication, creativity, and strong interest in AI/ML. Pathik completed coursework in Theory of Computation, Computer Networks, and Web Technology, built a JavaScript/CSS guitar-selling website, and worked on projects involving fake news detection and sentiment/performance analysis.”
“Recognized as a versatile, energetic, and out-of-the-box thinker with strong creativity in technical problem-solving. Pathik built a lightweight serverless video conferencing app using WebRTC and ICE candidates, showing initiative during online classes and successfully completing the project.”
“Recognized by PDEU for successfully completing his term as Event Management Head of Respawn, the university’s Gaming Club under OSAIL. Pathik was appreciated for his dedication, hard work, commitment, and sincerity in carrying out his leadership responsibilities during the 2021–22 academic year.”
Certifications & Papers
Proof of learning, practice, and technical depth.
A growing collection of verified certifications, public profiles, and papers.
Google Skills Profile
Verified Google Skills public profile showcasing completed learning paths and cloud/data skill development.
HackerRank Certification
Verified HackerRank certificate demonstrating practical coding and problem-solving ability.
Research Papers
Add published papers, preprints, technical blogs, or case studies here as the portfolio grows.
Contact
Let’s build something useful.
Open to software engineering, AI/ML, data engineering, and full-stack product roles.