Ahmedabad, India · Open to opportunities

Pratham Shah.
Backend Developer

I build Python backends, REST APIs, and multi-tenant SaaS systems with Django, FastAPI, and PostgreSQL, integrating third-party platforms, enterprise systems, and OpenAI-powered LLM workflows into production applications.

View my work Contact me

0

Database tables synced via REST APIs

0

Stakeholder personas in LLM reports

0

Automated tests on my resume matcher

01 · About me

A backend engineer who ships across the stack — and into the LLM layer.

I'm a backend-focused software engineer with a BE in Artificial Intelligence & Machine Learning and hands-on experience building and maintaining Python backends, REST APIs, and multi-tenant SaaS systems in production. At Akkomplish Group, I work across backend and frontend on BAS TrustDesk, where I integrated OpenAI-powered LLM analysis into backend workflows.

My range covers the whole product: Django / DRF / FastAPI backends, PostgreSQL data layers, React / Next.js / TypeScript frontends, and the AI layer on top: OpenAI API, LangChain / LangGraph, RAG, MCP, and tool-calling agents. I've integrated platforms and enterprise systems such as GitHub, Azure DevOps, and Microsoft Dynamics 365 Business Central.

Most recently I built Aster, a multi-user AI personal assistant that connects Gmail, Google Calendar, Outlook, and WhatsApp, turns live account data into AI-generated daily briefs, and chats through a streaming agent over official remote MCP servers.

Currently

Junior Software Engineer @ Akkomplish Group

Django REST · React 19 + TypeScript · OpenAI

Focus

Backend & full-stack engineering with AI/LLM

REST APIs · Multi-tenant SaaS · RBAC · LLM workflows

Certified

Microsoft Certified · DP-800 & DP-700

SQL AI Developer · Fabric Data Engineer

02 · Skills

Tools I think in.

03 · Experience

Where I've shipped.

Junior Software Engineer

Jun 2025 — Present

Akkomplish Group · Ahmedabad

  • Working across backend and frontend on BAS TrustDesk (Django REST Framework, React 19 + TypeScript), building modules across strategy assessment, JTBD-to-work-item alignment scoring, connector integrations, financing/roadmap tracking, and RBAC-gated stakeholder reporting dashboards.
  • Integrated OpenAI-powered LLM analysis into backend workflows to generate automated, role-aware strategy-to-execution alignment reports across nine stakeholder personas.
  • Developed CousinMiddleware, a middleware service bridging a database/Azure Functions layer with Microsoft Dynamics 365 Business Central, implementing List, Update, and Sync REST APIs across 61 database tables to automate CRUD-based data synchronization.
  • Owned delivered modules end to end: designing database schema changes and RBAC rules for system-wide impact, writing tests, debugging across the stack, and factoring shared API and UI logic into reusable components to cut rework on subsequent features.

Software Developer Intern

Jan 2025 — May 2025

Elixir Techne · Ahmedabad

  • Built REST API workflows and reusable backend service components for production web applications.
  • Optimized request handling in collaboration with frontend developers to improve response times and reliability.

04 · Projects

Selected work.

Featured · Flagship build

Node.js TypeScript Next.js 16 React 19 PostgreSQL MCP

Aster — Agentic AI Personal Assistant

A multi-user AI assistant on a Node.js/TypeScript backend that connects four providers per user, Gmail, Google Calendar, Outlook, and WhatsApp, and turns live account data into AI-generated daily and user-scheduled briefings. It integrates four official remote MCP servers from Google and Microsoft over a shared streamable-HTTP JSON-RPC transport, discovering tool catalogs at runtime via tools/list instead of hardcoding them.

  • Agentic chat: a bounded multi-step tool-calling loop over namespaced tools from all four providers, streamed over NDJSON
  • Per-user auth with server-side token storage: one shared Google grant with incremental authorization, and Microsoft Entra OAuth 2.0 with PKCE
  • WhatsApp via a Baileys pairing-code socket with DB-backed session state
  • MCP-first execution with a nine-tool REST fallback for ineligible accounts

Smart Resume Screening Tool

AI resume-to-job-description matcher

Ranks up to 20 resumes against a job description by meaning rather than keyword overlap, using all-MiniLM-L6-v2 384-dimensional embeddings and cosine similarity to output ranked candidates with match percentages.

  • Four-module NLP pipeline: validated PDF/DOCX/TXT parser, token-aware chunking, cosine similarity mapper, and a deterministic ranker
  • POST /api/match/ with a standardized response envelope, global exception handler, and 5 MB upload limits
  • Covered by 81 automated tests
Django REST Framework sentence-transformers Scikit-learn

Movie Recommendation System

Content-based recommender

Returns the 5 most similar movies for any title in the TMDB 5000 dataset, using stemmed bag-of-words vectors (CountVectorizer, 5,000 features) and a precomputed cosine similarity matrix serialized for fast inference.

Python Pandas NLTK Scikit-learn

05 · Education

Foundation.

LJ University

2021 — 2025

B.E. — Artificial Intelligence & Machine Learning

Coursework: Data Structures & Algorithms, Machine Learning, Neural Networks, DBMS, Software Engineering.

07 · Achievements

Numbers that shipped.

0

Database tables synchronized with Dynamics 365 Business Central through CousinMiddleware's List, Update, and Sync APIs.

0

Stakeholder personas served by automated, role-aware LLM alignment reports in BAS TrustDesk.

0

Automated tests covering the Smart Resume Screening Tool's NLP pipeline and API.

0

Official remote MCP servers from Google and Microsoft integrated into Aster.

08 · Contact

Let's build something intelligent together.

I'm open to backend, full-stack, and AI/LLM engineering roles or a conversation about an idea you want shipped. Reach out any way you like.

Email me Download Resume
prathamshah8487@gmail.com +91 84870 17317 prathamshah.me github.com/Pratham8487 linkedin.com/in/pratham1212