Shivam Raghav

Innovative and results-driven Python Developer and Data Scientist with a strong background in building and deploying AI solutions using FastAPI, Flask, and advanced NLP libraries such as Hugging Face Transformers. Skilled in developing deep learning models with TensorFlow and implementing computer vision systems using YOLO and Mediapipe. Proven expertise in designing scalable machine learning pipelines, optimizing workflows, and delivering high-impact AI projects across cloud platforms. Adept at cross-functional collaboration, with a strong focus on problem-solving, adaptability, and delivering robust, data-driven solutions that align with business goals.

• Description: Developed a privacy-focused conversational AI using OpenSource LLM models and integrated it with LangChain and LangGraph technology for secure data interactions.
• Tech Stack: Python, LangChain, GPT, Blockchain, FastAPI
• Impact: Enabled secure document querying and interaction while preserving user confidentiality.

• Description: Built an NLP application that extracts and summarizes content from uploaded PDF files using Transformer models.
• Tech Stack: Python, Hugging Face Transformers, PyPDF2, Streamlit
• Features: Keypoint extraction, abstractive summarization, and named entity recognition.

• Description: Created an intelligent chatbot capable of answering employee FAQs using RAG (Retrieval-Augmented Generation) with vector-based document search. It has automated pipelines to fetch the data from HR Policy Documents, IT Helpdesk, Finance Data
• Tech Stack: LangChain, OpenAI, ChromaDB, FastAPI
• Use Case: Streamlined internal HR communication and reduced support ticket load.

• Description: Developed a computer vision model using YOLOv8 and OpenCV to segment and analyze facial regions like Inner Canthus and under-eye areas for PTSD and TBI detection.
• Tech Stack: YOLOv8, OpenCV, Python
• Outcome: Delivered a proof-of-concept tool for early screening of mental health conditions.

• Description: Built a system to audit and score customer service and sales calls based on adherence to script, coherence, tone, and engagement.
• Tech Stack: Python, Speech-to-Text, NLP, Transformers
• Use Case: Improved quality control and performance evaluation of call center agents.

• Tech Stack: Python, FastAPI, PostgreSQL, REST APIs
• Description: Designed and implemented an automated scheduling system for workplace TV content using Gracenote metadata and BYOD GameMaster APIs.
• Key Features:
  - Integrated data from lineups, screen groups, and user preferences to create personalized daily and multi-day TV planners.
  - Fetched channel lineups via BYOD GameMaster API and retrieved program metadata from Gracenote.
  - Applied dynamic scheduling logic, including:
  - Prioritization of DirecTV for ESPN+ content.
  - Intelligent gap-filling with sports or fallback content.
  - Planned future integration of Trivia Channels.
  - Stored schedule outputs in PostgreSQL and exposed via RESTful API for frontend integration.

• Role: Python Developer | AI/NLP Engineer
• Tech Stack: Python, FastAPI, IBM Watson (LLM & NLP), Streamlit, PostgreSQL, VoiceBots, JWT, IBM Cloud
• Description: Developed a comprehensive hiring automation platform that leverages NLP and AI to streamline job description creation, candidate evaluation, and voicebot-led interviewing.
• Key Highlights:
  - NLP-based JD and CV generation using IBM Watson LLMs.
  - CV and JD semantic matching with NLP scoring and candidate classification.
  - Dynamic question generation for interviews with domain and difficulty tuning.
  - Automated interview scheduling and voicebot-led HR/technical assessments.
• Deployment: Fully deployed on IBM Cloud with Streamlit dashboard and PowerShell-based auto-launch setup.
• Impact: Reduced hiring effort by 80%+, personalized candidate experience, and enhanced recruiter efficiency with AI/NLP-driven decision support.

• Description: Built a classification pipeline to predict churn probability and segment employees into risk categories.
• Tech Stack: Scikit-learn, XGBoost, Pandas, Python

• Description: Created models to forecast future sales trends using time series modeling, lag features, and boosting algorithms.
• Tech Stack: LightGBM, ARIMA, Python

• Extramiler Award – Jun 2024, Netsmartz Infotech
  Recognized for exceptional dedication by working beyond standard hours to deliver a critical project ahead of schedule. Led the development and deployment of a Multi-Tenant AI Tools Platform with outstanding efficiency.
• Team Extramiler Award – Awarded Twice
  Honored as part of a high-performing team for collaborative excellence and delivering impactful AI-driven solutions under tight deadlines.
• Star Performer of the Quarter – 2-Time Recipient
  Selected for consistently exceeding performance metrics, demonstrating leadership, and contributing to key AI and automation initiatives.