Jethva
Parthiv
I design and build scalable agentic AI systems, multi-agent workflows, RAG pipelines, and AI-powered automation tools - prototyping fast and iterating based on real usage.
About Me
I'm an Agentic AI Engineer and Python Developer pursuing a Master's degree, with hands-on experience designing and building scalable agentic AI systems, multi-agent workflows, retrieval-augmented generation (RAG) pipelines, and AI-powered automation tools.
Experienced in developing agentic applications and scalable backend services using Python, FastAPI, LangGraph, LangChain, LangSmith, vector databases, and leading LLM APIs. I prototype fast with AI-assisted tools, iterate based on real usage, and focus on tools that automate workflows and save measurable time.
Areas of Interest
Python Ecosystem
Leveraging Python as a core language for versatile application development, system scripting, automation, backend services, and AI logic integrations.
Agentic AI & RAG
Orchestrating autonomous multi-agent workflows (LangGraph/LangChain) and advanced Retrieval-Augmented Generation (RAG) pipelines.
Machine Learning
Exploring supervised & unsupervised learning algorithms, classification, regression, and model evaluations with popular ML libraries.
Data Science
Uncovering patterns in data through statistical modeling, exploratory data analysis, feature scaling, and interactive visualizations.
Technical Skills
Languages & Concepts
LLM Harness
AI Concepts
Frameworks & Tools
Libraries
Databases
DevOps & LLMOps
Tools
Soft Skills
Projects
View all on GitHubOpsPilot
Autonomous RevOps AI Agent Platform
Architected an autonomous ReAct AI agent using LangGraph and LangChain — integrating Salesforce, Slack, Notion, and ChromaDB-powered RAG into a real-time AI operations platform.
- ReAct agent dynamically reasons, invokes tools, queries Salesforce CRM & retrieves knowledge from Notion via ChromaDB RAG
- Real-time event-driven platform using Slack Socket Mode + FastAPI backend + React dashboard with SSE
- Human-in-the-Loop (HITL) approval system for reviewing AI-generated responses before publishing to Slack
- Telemetry for token usage, latency & cost tracking with interactive ROI and automation dashboards
NoteFlow
Open SourceAutonomous voice-driven Notion assistant using FastAPI, LangGraph, and Google Gemini connected to Notion MCP — transcribing voice via Groq Whisper and delivering spoken confirmations via edge-tts.
- Cut capture latency by over 70% by pairing an in-memory workspace cache with prompt-tuned single-turn creation for new notes, while maintaining full multi-turn ReAct loops for exploratory multi-step page updates
- Built LangGraph duplicate-tool interception and recursion circuit breakers to stop infinite search loops
- Implemented English-locked, zero-temperature STT to reduce hallucinations, and used LangSmith to trace agent trajectories, tool calls, and token usage
RAGVerse AI
ImprovingDesigned and implemented a modular Retrieval-Augmented Generation (RAG) system using LangGraph and LangChain, enabling multi-turn conversational interactions with persistent memory.
- Modular RAG system enabling multi-turn conversational interactions with persistent memory checkpointing in PostgreSQL
- ChromaDB retrieval pipeline with hybrid retrieval, reranking, and configurable document chunking
- RAGAS-based evaluation pipeline measuring answer faithfulness, answer relevancy, context precision, and retrieval quality
- Extensible design supporting future LLM guardrails, query rewriting, contextual compression, and adaptive retrieval
ResearchFlow
LiveAutonomous deep research agent powered by LangGraph and Gemini LLM — orchestrating iterative self-correction loops and concurrent claim-level fact-verification against live web sources.
- Iterative self-correction research loop that evaluates evidence and refines search queries dynamically
- Concurrent claim-level verification pipeline that fact-checks findings against live web sources
- Persistent Streamlit dashboard with LangSmith observability and tracing of multi-agent graphs
My GitHub MCP Server
LiveProduction-ready Model Context Protocol (MCP) server built in Python using FastMCP — serving as the foundational toolset layer for autonomous AI PR review agents.
- Implemented 7 specialized PR tools covering metadata extraction, unified diffs, changed files, and inline code reviews with line targeting
- Accepts repository identifiers and auth tokens as dynamic runtime parameters to inspect public and private repos without reconfiguration
- Modular, extensible architecture designed to expand into custom GitHub workflows with resilient error handling
- Deployed on Render with Streamable HTTP transport, JSON-RPC communication, and Pydantic validation
AI Event Scout Agent
ImprovingAutonomous multi-agent system that discovers, validates, deduplicates, and ranks AI events from the web — replacing a fully manual research process.
- Agent workflows covering web search, scraping, structured extraction, and duplicate detection
- Exposed as a FastAPI service for integration with downstream tools
- Modular architecture allows new data sources to be added in under 30 minutes
Dynamic Filter System
Open SourceProduction-ready FastAPI backend for enterprise datasets supporting dynamic filtering, sorting, pagination, and multi-attribute querying via REST APIs.
- Dynamic filtering, sorting, pagination, and multi-attribute querying via REST APIs
- Optimized PostgreSQL queries through indexing and efficient SQL execution
- Significantly improved API response times for large datasets
Achievements & Certifications
Verify Certificates1st Place — Python Programming Competition
2026Secured 1st place in the university-level Python programming competition, solving advanced logic, data structures, and algorithmic problems.
1st Place — Code Siege (Python)
Feb 2026Secured 1st place in Code Siege. A unique two-phase battle: I built the initial Python solution, then an opponent deliberately injected bugs — and my teammate, working in complete isolation without the problem statement, had to understand the codebase and restore it under time pressure.
Runner-up — Logic Forge (Python)
Feb 2026Won second place in the university-level Python coding marathon, designing algorithms to solve complex logic challenges under strict run-time optimizations.
1st Place — Code Combat (Java)
Jan 2026Ranked 1st in Code Combat (PG Level – Java) on January 23. Started from an elimination round with 45 teams and worked through multiple coding challenges under time pressure with a clear strategy.
Participant — n8n Community Hackathon
Aug 2026Demonstrated innovation and technical skills in developing AI-powered workflow automation solutions using the n8n platform, designing robust automated pipelines and AI integrations.
Participant — Supervity AutoPilot Hackathon
2026Successfully participated in the Supervity AutoPilot Ahmedabad Hackathon, building an AI Employee that pushed the boundaries of workflow automation.
Participant — SSIP Regional Hackathon
2023Built FitAthlete — a smart nutrition & diet recommendation web app for athletes — during a 24-hour hackathon. Tracked calories, fats, carbs and suggested nutritious dishes with exact nutritional breakdowns. Built with Python (Flask), HTML, CSS & JavaScript.
AWS Generative AI: Art of the Possible
OnlineCompleted training on the fundamentals of Generative AI, foundational models, use cases, and deployment strategies in the AWS ecosystem.
GitHub
@Jethva-ParthivAutonomous RevOps AI agent platform — LangGraph + Gemini + Salesforce + Slack + Notion.
Autonomous voice-to-Notion capture agent — FastAPI + LangGraph + Gemini + Notion MCP + Groq Whisper + edge-tts.
End-to-end modular RAG system using LangGraph, LangChain, ChromaDB, and PostgreSQL.
Autonomous deep research agent with iterative self-correction and live fact-checking.
Multi-agent system that discovers, validates, deduplicates, and ranks AI events from the web.
Model Context Protocol (MCP) server enabling AI agents and LLMs to interact seamlessly with GitHub APIs.
Production-ready FastAPI backend for enterprise datasets with dynamic filtering and pagination.
Get In Touch
Interested in working together? Whether it's a freelance AI project, a full-time role, or just a chat about agentic AI — I'd love to hear from you.