I'm a final-year Computer Engineering student focused on building scalable backend systems, high-throughput APIs, and production AI architectures.

Most of my engineering work is centered on building reliable software from the ground up — designing low-latency REST APIs, structuring clean PostgreSQL schemas, and writing defensive backend logic that handles edge cases gracefully.

I've taken three projects from whiteboard concept to production: CampusCircle (a verified community and learning platform I built solo on FastAPI and PostgreSQL, scaled to 200+ active users), AgentPay (AI buyer-agent commerce infrastructure for the Razorpay Buildathon, using an isolated policy engine to strictly gate financial actions), and CliniqueAI (combining ML classification models with clinical document retrieval).

Outside of projects, I have competed in 6 hackathons, placing in the Top 20 out of 1,600+ at Unstop's National AI Hackathon and Top 18 out of 500+ at The Maverick AI Challenge. I'm currently looking for software engineering, backend, and applied AI roles where I can build and maintain serious systems.

Projects

Production Systems & R&D

Python · FastAPI · LangGraph · PostgreSQL · Redis · Next.js · LLaMA 3.3

AI Agents LangGraph FastAPI FinTech Policy PostgreSQL Redis
  • Built autonomous buyer-agent infrastructure where the LLM proposes purchases, but an isolated deterministic policy engine gates every payment action before Razorpay’s API is called — the agent never directly initiates money movement.
  • Enforced an append-only, database-trigger-backed audit trail logging every state transition, tenant merchant isolation, and hard consumer spend-authorization limits.
  • Shipped a full sandbox-to-live compliance pipeline featuring webhook-driven external agent integration with HMAC verification and automatic retry handling.

Next.js · FastAPI · PostgreSQL · LLaMA 3.3 70B · TypeScript

Community 200+ Users FastAPI LLaMA 3.3 70B PostgreSQL
  • Architected the complete backend and AI layer for a verified campus community platform, solo developer scaling to 200+ active students.
  • Shipped 3 AI features powered by LLaMA 3.3 70B: student support assistant ("Reva"), an adaptive quiz engine adjusting difficulty dynamically, and a YouTube-to-structured-notes generator.
  • Placed in the Top 18 out of 500+ participants in The Maverick Ect AI Challenge 2026 (Dewang Mehta Foundation Trust).

CliniqueAI — AI-Powered Clinical Decision Support

Top 20 of 1,600+ · Unstop Hackathon

Python · Flask · PostgreSQL · Scikit-learn · XGBoost · LangChain · Pydantic

Healthcare ML RAG XGBoost Scikit-learn Flask PostgreSQL
  • Trained and benchmarked Logistic Regression, Random Forest, and XGBoost classifiers for diabetes risk prediction, selecting optimal thresholds for clinical explainability.
  • Implemented a RAG patient assistant grounded on localized EHR history, complete with transactional CRUD operations, Pydantic validation, and physician notification alerts.
  • Eliminated free-tier cloud cold starts on Render using an automated 10-minute heartbeat via UptimeRobot.

Experience & Education

Timeline
1-Page PDF ↗
B.Tech in Computer Engineering 2023 — 2027
Sankalchand Patel University · Visnagar, Gujarat
CGPA: 7.80 / 10.0
  • Coursework: Machine Learning, Artificial Intelligence, Deep Learning Applications, Database Management Systems, Operating Systems, Data Structures & Algorithms, Software Engineering, NLP, Compiler Design, GenAI.
  • Organizer in the college technical club — managing hackathons and technical workshops.
Top 20 Finalist (out of 1,600+ Participants) Feb 2026
National AI Hackathon · Unstop
Verified national finalist for developing CliniqueAI, integrating predictive ML diagnostic classification with a RAG patient chatbot.
Top 18 Finalist (out of 500+ Participants) 2026
The Maverick Ect AI Challenge · Dewang Mehta Foundation Trust
Recognized for CampusCircle, scaling an AI-augmented community hub to 200+ active students using LLaMA 3.3.
Hackathon Competitor (6 Total) Aug 2026 — Sep 2026
Razorpay Buildathon, Unstop, DMF Trust Hackathons
Built end-to-end full-stack AI and backend systems under 24–48 hour constraints, focusing on agent workflows, deterministic validation, and transactional data integrity.
Certifications & Foundations 2025 — 2026
HackerRank · Edunet Foundation
  • HackerRank Certified Software Engineer: Problem Solving, SQL, REST API Design (2025).
  • Foundation Course on Green Skills & AI: Edunet Foundation (2026).
  • Operating Systems & Python Programming Foundations.

Technical Skills

Proficiencies & Architecture

Languages

Python C++ SQL TypeScript

Backend & APIs

FastAPI Flask RESTful APIs Async I/O Pydantic Webhooks HMAC Verification Rate Limiting

AI & Agentic Systems

LangGraph LangChain Claude API OpenAI API RAG Pipelines Tool/Function Calling Prompt Engineering

Databases & Vector Search

PostgreSQL Redis SQLite FAISS Qdrant Pinecone

System Design & Scalability

Docker Microservices Message Queues (Celery) Load Balancing Caching Strategies API Gateways CI/CD Nginx

DevOps & Tools

Git GitHub Linux/Bash Vercel Render Postman

Writing & Logs

Reflections on artificial general intelligence, research methodology, and machine learning architectures.

Contact

I'm actively looking for opportunities in backend software engineering, scalable systems, and applied AI. If you're building something interesting or want to collaborate, feel free to reach out:

Email: prempatel7740@gmail.com
LinkedIn: linkedin.com/in/prempatel-ai
GitHub: github.com/prempatel-ai
Medium: medium.com/@prempatel7740