Developer

About Me

Get to know me better

I am an AI Engineer and Full-Stack Developer passionate about turning ambitious ideas into working products. Over the last year, I have built 40+ AI Multi-Agent Systems and Full-Stack applications and shipped them on free platforms like Vercel, Render, Railway, and Hugging Face, taking ideas from scratch to production-ready MVPs. I love designing intelligent systems that do not just demonstrate AI capabilities, they solve real business problems, automate complex workflows, and create measurable impact.

My work spans LLMs, Multi-Agent Architectures, RAG, AI Automation, Backend Systems, and Modern Full-Stack Development. Whether it is building autonomous AI workforces, developer tools, or enterprise automation platforms, I enjoy owning the complete engineering journey from architecture and backend infrastructure to AI orchestration and deployment.

I am driven by curiosity and obsessed with building. Every project is an opportunity to learn something new, experiment with emerging technologies, and transform an idea into something people can actually use.

When I am not coding, you will usually find me exploring cutting-edge AI research, contributing to open-source projects, or building the next product that pushes the boundaries of what AI can accomplish.

Skills

The tools and frameworks I use to bring ideas to life

Languages

Python
TypeScript
JavaScript
C++

AI & LLM Engineering

LangGraph
LangChain
RAG Pipelines
Multi-Agent Systems
LLM Integrations
Prompt Engineering
AI Workflow Automation

Full Stack Web Development

HTML
CSS
React.js
Next.js
Node.js
Express.js
Supabase
JWT Token
REST API / Microservices
Kafka
RabbitMQ
FastAPI

Databases

PostgreSQL
MongoDB
Pinecone
FAISS
ChromaDB

Tools & Deployment

Git/GitHub
Linux
AWS EC2
AWS S3
Vercel
Render
Railway
Hugging Face
Docker
Kubernetes

Featured Projects

Some of the things I've built recently

OmniForce — Autonomous Enterprise AI Workforce
OmniForce — Autonomous Enterprise AI Workforce
Multi-agent AI platform automating KYC/AML compliance and workflows using LangGraph, ChromaDB RAG, and LLaMA 3.3 70B inference.
PatentPilot AI — Multi-Agent Patent Analysis
PatentPilot AI — Multi-Agent Patent Analysis
Architected a 6-agent parallel reasoning system for automated patent analysis using FAISS vector indexing and Groq-accelerated LLaMA 3.1.
SQL Editor with AI Assistance
SQL Editor with AI Assistance
Production-grade SQL sandbox used by 200+ users for real-time query execution, automated grading, and interactive practice with Gemini LLM hints.
Scientific Paper Knowledge Extractor
Scientific Paper Knowledge Extractor
Scalable LLM-powered pipeline extracting structured JSON from 200M+ scientific papers using Groq-accelerated LLaMA 3.1 and Semantic Scholar API.
Infrastructure Observability & Alerting Stack
Infrastructure Observability & Alerting Stack
Real-time metrics monitoring with Prometheus and Grafana, featuring Alertmanager routing and a Python-based SSH brute-force detection script.
Legal & Tax RAG System
Legal & Tax RAG System
Enterprise-grade AI research assistant for US Tax Law. Built with Pinecone, Graph RAG (NetworkX), and Groq LLaMA 3.3 for grounded citations.

Best Open Source Contribution

Making impactful contributions to widely-used AI frameworks

DSPy — Stanford NLP

34K+ GitHub Stars

Contributed a production-grade fix to DSPy, Stanford NLP's declarative LLM framework — resolving a critical silent state-corruption bug in load_state() causing inconsistent RAG and inference behavior.

Implemented a validate-first + deepcopy transactional loading strategy ensuring all-or-nothing state restoration, analogous to database BEGIN/ROLLBACK/COMMIT semantics — shipped with 47 passing tests.

Experience & Education

My professional journey

ISRO
ISRO
June 2025 - Aug 2025

Indian Space Research Organization (ISRO)

ML Intern

Built end-to-end 6-DoF spacecraft pose estimation pipeline, achieving ER 3.97° and ET 2.26m on a 12K+ dataset. Engineered dual-head CNN architecture for orientation and translation prediction.

Python
TensorFlow
PyTorch
Computer Vision
DTU
DTU
May 2024 - July 2024

DTU USIP Program

Software Engineering Intern

Built production-grade Hostel Management System used by 2,000+ students. Architected backend with Node.js, MongoDB, and JWT-based authentication with RBAC. Developed full-stack system with Next.js.

Node.js
MongoDB
Next.js
React
DTU
DTU
Nov 2022 - Jun 2026

Delhi Technological University

B.Tech in Computer Science Engineering

Current CGPA: 7.65. Relevant Coursework: Data Structures, Algorithms, Web Development, and Software Engineering.

C++
Git
JavaScript

Get In Touch

Have a project in mind or just want to say hi?

Send me a message
Or email me directly at soniashish37066@gmail.com