SHARVIN VARGHESE
+91-971******* ******************@*****.*** Noida and Gurgaon, India (open to remote) linkedin.com/in/sharvin-varghese-66b04917a github.com/sharvinvarghese Professional Summary
Generative AI engineer with eight months of hands-on LLM application development at Tooliqa, promoted from Data Science Intern to Junior Data Scientist. Built OpenAI tool-calling chatbots over structured survey data, embedding-based semantic search (Gemini API, all-MiniLM-L6-v2), multi-agent financial monitoring workflows, and speech-to-LLM interview analysis pipelines, supported by Apache Airflow ETL with Pydantic validation. Also built a 10-node LangGraph agent with model fallbacks and failure recovery. Prior computer vision experience
(YOLO) at Helmerich & Payne.
Skills
Programming: Python, SQL
Generative AI and LLMs: OpenAI API, Gemini API, Groq API, Transformers, Qwen, vLLM, prompt engineering, structured outputs, model evaluation, speech APIs (Google Speech-to-Text, AssemblyAI, Sarvam AI, Speechmatics)
RAG and Retrieval: embeddings (all-MiniLM-L6-v2), semantic search, vector search, retrieval pipelines Agentic AI: LangGraph, LangChain, tool calling, multi-agent systems, MCP; AI-assisted development with Claude Code and OpenAI Codex
Data Engineering and MLOps: Apache Airflow, ETL, Pydantic, API integration (Polygon.io), Docker, CI/CD ML and Computer Vision: PyTorch, TensorFlow, Scikit-learn, YOLO, OpenCV, SparseML, data annotation Cloud and Tools: AWS, Azure, Django, Streamlit, Git Work Experience
Tooliqa Innovations LLP, Gurgaon Jan 2026 – Aug 2026 Junior Data Scientist (promoted from Data Science Intern) Jul 2026 – Aug 2026
• Built an OpenAI-powered chatbot with tool calling that queries, extracts, and processes tabular survey data from databases, integrating in-house table-extraction utilities for structured analysis.
• Designed multi-agent LLM workflows for financial monitoring that ingest and analyze Polygon.io market data at 5-minute intervals.
• Built an interview intelligence pipeline combining speech-to-text, text-to-speech, and LLMs to transcribe interviews, summarize them, and extract themes; evaluated and integrated Google Speech-to-Text, AssemblyAI, Sarvam AI, and Speechmatics.
Data Science Intern Jan 2026 – Jun 2026
• Built a semantic-search chatbot over tabular datasets using the Gemini API and all-MiniLM-L6-v2 embeddings, enabling natural-language querying.
• Combined embedding-based retrieval with tool calling to send only relevant data to the LLM, reducing unnecessary context and generating structured analytical reports.
• Built Apache Airflow ETL pipelines with Pydantic validation to automate financial data processing, insight generation, and reporting.
Helmerich & Payne Dec 2023 – Apr 2025
ML-CV Trainee
• Built YOLO-based computer vision models for pipe height estimation and perimeter-breach detection, improving detection accuracy by 25% and reducing inspection and safety-monitoring time by 40%.
• Integrated a Microsoft T5-based code reviewer via API to evaluate interview and production-level code, delivering structured code-quality feedback through automated review workflows. Cognizant Oct 2023 – Dec 2023
Programmer Analyst Trainee
• Debugged and maintained enterprise mainframe applications using JCL, COBOL, and ISPF, collaborating with cross-functional teams to troubleshoot and resolve application issues. Amity University Jan 2023 – May 2023
Machine Learning Intern
• Developed a Django-based Hospital Management System with backend workflows and database integration for hospital records, adding machine learning and analytics features for clinical and operational use cases. Xcitedu Software Solutions Feb 2021 – Jun 2021
Machine Learning Intern
• Developed Django and Python web applications and REST APIs for image-processing-based apparel recommendations, and built a prototype chatbot using deep learning and NLP techniques. Blue Planet Infosolutions Oct 2020 – Jan 2021
Machine Learning Intern
• Developed a Django and MySQL Diet Diary application for tracking dietary information, including backend functionality and database integration for user records. Projects
AI Newsletter Agent github.com/sharvinvarghese/ai-newsletter-agent Python, LangGraph, Groq, Pydantic, Streamlit
• Built an end-to-end newsletter generation agent as a 10-node LangGraph workflow covering goal parsing, research planning, multi-source retrieval, article evaluation, summarization, generation, critique, and persistence.
• Implemented multi-model fallbacks, Pydantic schema validation, rate-limit handling, and rule-based degradation so runs continue when LLM calls fail.
• Developed a Streamlit interface with live workflow progress, and saved each run’s artifacts (JSON, Excel, summaries) for replay, debugging, and newsletter download. Chess AI with SLM github.com/sharvinvarghese/chess-ai-slm Python, Gemini API
• Built a browser chess game with a Gemini-powered AI opponent, real-time in-game chat, and LLM-generated move hints triggered by a keyboard shortcut.
Label Converter github.com/sharvinvarghese/label-converter Python, YOLO, COCO, Pascal VOC
• Developed a zero-config CLI and Python API that converts object detection annotations between YOLO, COCO, and Pascal VOC formats with a single command. Education
Amity University 2018 – 2023
Integrated B.Tech + M.Tech in Biotechnology
Publication
“Precision at Its Core: Machine Learning Infused Metabolomics Model for Preterm Birth Prediction in Humans,” IEEE Xplore, document 10307581.