Vishnu Singh
9530396291 — # singhvishnu9530@gmail.com — ï linkedin.com/in/vishnu-singh-85277419b/ — § github.com/vishnu020
Summary — Data Scientist with 2.5 years of experience in ML, NLP, and Gen-AI. Proven expertise in time series forecasting,
LLM observability, RAG-based retrieval systems, and agentic AI implementations. Strong background in building scalable ML
solutions for business applications using Python, SQL, and cloud computing platforms. Experience in developing automated,
intelligent systems that optimize business processes and drive measurable impact.
Skills
– ML & AI: Supervised & Unsupervised Learning, Time – LLMs & RAG: Adaptive RAG, Gemini-Pro, GPT-4, GPT-3.5,
Series Forecasting, Deep Learning, Gen-AI, Transformers, Cohere, OpenAI, Gemma, Llama, Mistral
Statistics, Predictive Analytics, CV, Neural Networks, LLMs – Databases: Neo4j, ArangoDB, FAISS, ChromaDB,
– Languages: Python, SQL, PySpark Pinecone, Weaviate
– Libraries: Pandas, NumPy, PyTorch, TensorFlow, – MLOps: MLflow, Arize AI, Docker
Scikit-Learn, Seaborn, Matplotlib – Big Data & Cloud: Databricks, Azure ML
– Frameworks: LangChain, LlamaIndex, Hugging Face, – Tools: Excel, Streamlit, OneReach AI, Power BI
NVIDIA NeMo, Azure OpenAI, OpenAI API
Experience
Visa Jun 2025 – Current
Data Scientist(Associate)
– Architected and implemented a multi-agent AI system using LangGraph that automatically generates production-ready
UI components from Figma designs, reducing development time by 70
– Developed a sophisticated workflow orchestration system with specialized agents (Analysis, Developer, Validation) that
handle different stages of the UI component generation process with iterative refinement capabilities
– Created an AI-powered design analysis pipeline that extracts and interprets complex Figma design specifications,
supporting responsive layouts with desktop and mobile variants.
– Built a robust feedback loop system that enables iterative refinement of generated components based on validation
results and user feedback, ensuring high-quality output.
– Developed end-to-end AI-powered video generation system that converts web content into professional videos using
Google’s Gemini AI and Veo3 models, with intelligent content processing pipeline featuring web scraping, HTML
parsing, and automated image extraction using BeautifulSoup
Tredence Analytics Jun 2023 – June 2025
Data Scientist(Analyst)
– Led the implementation of Facebook’s Prophet for sales forecasting, achieving MAPE 5% across eight time series,
demonstrating expertise in predictive analytics and data-driven decision making.
– Created Microeconomics features for the Milky Project at the department level, independent of the Microeconomics
features provided by Walmart, by conducting research using articles and reports
– Developed an LLM observability solution for Pepgenx, reducing model failure incidents by 20% through Arize-based
anomaly detection and performance monitoring
– Optimized the PGT chatbot pipeline by integrating Arize, improving response tracking efficiency by 30% and reducing
debugging time by 40% through automated trace analysis
– Built an AI-driven bot leveraging advanced NLP, improving code retrieval accuracy by 40%, reducing manual search
time by 60%, and enhancing developer workflow. .
Nvidia Nims Hackathon July 2024–Oct 2024
Data Scientist(Analyst)
– Developed an AI-powered study assistant for NEET exam preparation, leveraging Llama 3.1 LLM to provide personalized
insights and responsive support to students. .
– Utilized Django for backend support, ChromeDB as a vector store for efficient query processing, and adaptive RAG
techniques for dynamic content delivery based on individual performance and question patterns.
Education
IIT DHANBAD
B.TECH-MINING MACHINERY ENGINEERING June 2019–May2023
Certifications
– Azure - Microsoft Azure Ai Engineer Associate
– Oracle - Oracle Cloud Infrastructure 2024 Generative AI Certified Professional
– Nvidia- NVIDIA Certified Associate - Generative AI LLMs
