AI Engineer who builds things that actually work.
I spend most of my time building production ML systems and RAG pipelines. Really into multi-agent orchestration with LangChain, making vector databases fast, and figuring out how to make LLMs reliable enough for real use cases.
Also mess around with graph neural networks, distributed systems, PySpark pipelines, microservices architecture, and whatever interesting ML problem shows up. Built everything from COBOL-to-microservice migration tools to multimodal AI systems.
Currently exploring: LangGraph for complex agent workflows, Rust for systems programming, and optimizing vector search at scale.
Next.js TypeScript Python MCP Gemini
Resume management system with Git-based version control and AI-powered DOCX parsing. Integrates Model Context Protocol for document processing, uses Gemini 2.5 Flash for structured data extraction, and generates PDFs with Puppeteer. Built custom JSON-RPC bridge for Next.js to Python MCP communication.
Python LangGraph FastAPI Gemini WebSockets
Multi-agent code generation system using LangGraph orchestration. Four specialized agents (Planner, Architect, Coder, Validator) collaborate to generate production-ready code from natural language. Real-time WebSocket updates, automated quality validation, and stateful session management.
Python FastAPI RAG LangChain DuckDB Chroma
Enterprise AI assistant with hybrid retrieval - combines RAG for documents and natural language to SQL for analytics. GPT-4 query classification, Cohere reranking, role-based access control, and intelligent fallback mechanisms. Handles both structured (DuckDB) and unstructured (Chroma vector DB) data sources.
Python PyTorch Graph Neural Networks Drug Discovery
Two hypergraph neural network approaches for molecular toxicity prediction. K-clique model uses dense subgraphs with heterogeneous message passing, functional group model uses SMARTS pattern matching. Achieved 0.894 ROC AUC on Tox21 androgen receptor dataset. Includes complete research paper.
Python Transformers Streamlit NLP
Web-based sentiment analyzer using RoBERTa transformers with automatic language detection and model selection. Extracts readable content from URLs, performs sentence-level sentiment highlighting, and supports batch processing via CSV. Includes caching and multilingual support with XLM-RoBERTa.
C TCP/IP Sockets Systems Programming
Four-node distributed file system in pure C with custom TCP protocol. Each server manages independent namespace with upload, download, delete, and tar operations. Client routes requests to appropriate nodes. Built from scratch without frameworks to understand low-level networking.
Python Random Forest Isolation Forest Digital Twin
Hybrid ML system for lithium-ion battery recycling optimization. Predicts chemical extraction of lithium, nickel, and cobalt from black mass using Random Forest. Includes contamination detection with Isolation Forest, digital twin simulation for batch behavior, and metal price forecasting.
Django SQLite Bootstrap Full-Stack
Sustainability tracking platform with gamification. Users set environmental goals across categories (waste, energy, food, transport), earn eco-points, and engage with community through likes and comments. Includes blog CMS, user profiles with achievement badges, and comprehensive admin panel.
React Native Flask Raspberry Pi GrovePi IoT
IoT automation system using NFC tags to trigger GPIO devices on Raspberry Pi. Flask REST API receives NFC commands from mobile app, controls relays/LEDs via GrovePi. Includes mobile dashboard for manual control and device management. Built for accessibility and smart home use cases.
Python LSTM Time Series Deep Learning
LSTM neural network for financial time series forecasting. Jupyter notebook implementation with data preprocessing, model training, and prediction visualization. Explores sequence modeling for stock price movements.
Languages: Python, TypeScript, JavaScript, C#, SQL, C
GenAI/LLM: LangChain, LangSmith, AutoGen, CrewAI, OpenAI, Claude, Gemini, Llama, Hugging Face, Model Context Protocol
RAG/Vector: FAISS, Pinecone, Weaviate, BERT embeddings, semantic search
ML/Data: PyTorch, TensorFlow, scikit-learn, PySpark, Databricks, Kafka, pandas
Backend: FastAPI, Flask, Django, Next.js, Node.js, ASP.NET Core, GraphQL
Cloud/DevOps: AWS, Azure, GCP (Vertex AI), Docker, Kubernetes, FluxCD, GitHub Actions, Azure DevOps
Databases: PostgreSQL, MongoDB, Redis, Vector DBs
What I'm actually good at: Building production RAG systems, multi-agent orchestration, MLOps pipelines, microservices architecture, making LLMs reliable enough for enterprise use
Contact: nilkanthsuthar@protonmail.com | LinkedIn