Elambharathi Thangavel

Python Engineer · AI/LLM Applications · Bengaluru, India

Resume (PDF) · GitHub · LinkedIn · Email · +91 9677407190


About

Python Engineer building production-grade AI agents and LLM applications with LangGraph, RAG, and MCP. Built an autonomous Jira-to-PR coding agent with self-healing test loops, Playwright visual verification via MCP, and human-in-the-loop approval — with full LangFuse observability. Also built an agentic RAG chatbot with two-stage retrieval and config-driven architecture. Won PhonePe’s AI Immersion Hackathon. Backed by production experience — automation pipelines on HDFS, on-call across 6 microservices, and FastAPI backends with PostgreSQL.


Experience

Product Solution Engineer · PhonePe

Bengaluru · Jul 2025 – Present India’s leading digital payments platform — Quick-Commerce division (Pincode)

  • Won the PhonePe “AI Immersion Hackathon” — built a complete end-to-end appointment booking system (merchant-facing: salons, spas, etc.) in 5 days using AI-assisted development. Led architecture design, API contract definition, and backend integration; judged on production-readiness and code quality.
  • Built an end-to-end automation pipeline (Python + PySpark + Hive) replacing a paid third-party platform — extracts merchant data from HDFS, applies business logic, and creates tasks via internal REST APIs. Runs twice daily via cron with 99%+ uptime and zero manual intervention.
  • Provided L1/L2 on-call support across 6 distributed microservices (Catalogue, Transaction, Fulfillment, Search, Buyer/Seller Apps). Achieved best-in-team metrics: MTTA < 1hr, MTTR < 13hrs through systematic log analysis and cross-service debugging.
  • Served as primary technical POC for merchant API integrations (Order, Inventory, Catalog APIs) — led end-to-end lifecycle from consultation to production sign-off with 100% success rate. Debugged API request/response flows using Postman and log analysis tools.
  • Built an automated reporting pipeline: Google Sheets API → HTML templating → PDF generation → SMTP email delivery, eliminating daily manual reporting work.

Python Developer · Chatbot Engineer · Jobshine

Singapore (Hybrid) · May 2024 – May 2025 Singapore-based startup — blue-collar recruitment automation

  • Built and maintained AI chatbots across WhatsApp, Messenger, and Telegram using Python and Make (Integromat); automated candidate engagement workflows and job application pipelines for messaging platforms.
  • Developed in-house Python web crawlers (Selenium, BeautifulSoup, Requests) replacing a paid scraping tool, saving 100% in tool costs and reducing manual data collection by 70%.
  • Performed EDA on engagement data using Pandas to improve lead quality and recruitment conversion outcomes.

Jira Coding Agent — AI-Powered Code Automation

Apr 2026 · GitHub

Built an autonomous AI coding agent using LangGraph that watches Jira for tickets, modifies a React codebase via RAG-powered code search (FAISS + grep), runs tests with a self-healing retry loop, and creates PRs with before/after screenshots via Playwright MCP. Includes human-in-the-loop approval for high-risk changes and full LLM observability via LangFuse.

Tech: Python, LangGraph, FAISS, Playwright MCP, LangFuse, GitHub API, Jira API

Bubble — Production-Grade RAG Chatbot

Apr 2025 – Present · GitHub · Try it

Built an agentic RAG chatbot using LangChain, LangGraph, and FAISS with a two-stage retrieval pipeline: bi-encoder (FAISS) for fast recall → cross-encoder reranking for precision, with configurable relevance thresholds. Designed config-driven architecture with Pydantic validation and YAML config — switching LLM providers, embedding models, or retrieval parameters requires zero code changes. Improved chunking from 322 noisy fragments to 82 high-quality chunks.

Tech: Python, LangChain, LangGraph, FAISS, Sentence Transformers, Cross-Encoder Reranking, Pydantic, Streamlit

FinSight — Stock Portfolio Tracker & Sentiment Analyzer

May 2025 · GitHub

Full-stack application with FastAPI backend (async endpoints, JWT auth, dependency injection), PostgreSQL database (Neon), real-time P&L tracking via yFinance, and news sentiment analysis using FinBERT. Deployed with Streamlit UI.

Tech: Python, FastAPI, PostgreSQL (Neon), yFinance, FinBERT, Streamlit, JWT


Skills

AI / GenAI: LangChain, LangGraph, RAG, FAISS, Agentic AI, Sentence Transformers, Cross-Encoder Reranking, Prompt Engineering, LLM APIs (Groq, OpenAI), MCP

Languages: Python, SQL, Shell Scripting

Backend: FastAPI, REST API Design, Async Python, PostgreSQL, Redis, Pydantic, JWT Authentication, Rate Limiting

Data & Infra: PySpark, Apache Hive, HDFS, Airflow, Cron, Docker, Git, Postman


Education

B.Tech in Computer Science and Engineering · Amrita Vishwa Vidyapeetham University

Coimbatore, India · Aug 2020 – May 2024 · CGPA: 7.29


Certifications


Publications

A Multi-Round Zero Knowledge Proof Algorithm for Secure IoT and Blockchain Environments International Journal of Safety and Security Engineering


Beyond Code

I also write about books I read, places I travel, and lessons I’m picking up along the way. The rest of my digital garden is open if you’re curious.


Get in touch

Open to AI Engineer / Python / Backend roles. Happiest when there’s a hard problem and a small team.

Email me · LinkedIn

0 items under this folder.