Ujwal Tandon

Dynamic data science leader with extensive experience at Paxcom India, excelling in team mentorship and innovative problem-solving. Proficient in Python and machine learning, I drive impactful projects that enhance KPIs and optimize processes, fostering a culture of collaboration and continuous improvement. Recognized for delivering data-driven insights that significantly reduce costs.

Autonomous, intelligent systems that use LLMs and frameworks to reason, evaluate, and act across multiple steps:

• Analyze AI
  Built a LangGraph-powered platform for autonomous data analytics, generating insights from structured datasets and visualizing them interactively.
• Text2SQL Chatbot
  Developed an agentic chatbot using LLaMA, LangChain, and LangGraph to convert user queries into SQL and handle multi-turn interactions with memory.
• Contract Analyzer
  Created an LLM-driven system for compliance checks across legal contracts, extracting clauses, assigning risk, and validating against regulations.
• TruthTracker: AI-Powered Twitter Fact-Checking System
  Built a real-time fact-checking tool using Groq AI and multi-source verification, analyzing tweet credibility and engagement context with autonomous logic.
• Context Sufficiency Evaluator API
  Designed an API using Groq LLM to assess whether a provided context is sufficient to answer a question. Evaluates relevance, identifies gaps, scores confidence, and ensures quality of context prior to generation.

Focused on classical machine learning, computer vision, intelligent automation, and data pipelines:

• LinkedIn Connection Automation Tool
  Engineered a Selenium-based automation system for sending personalized connection requests on LinkedIn with anti-detection mechanisms and CSV bulk processing.
• Invoice Payment Automation
  Built a pipeline using YOLO, LayoutLM, and OCR to automate invoice parsing and payment processing by extracting essential financial information.
• Aspect-Based Sentiment Analysis
  Used BERT models to analyze sentiments across different product review aspects and detect misleading or fake profiles.
• Marketing Mix Modeling
  Developed a machine learning solution to analyze marketing channel effectiveness and attribute performance metrics.
• Image Similarity Comparator
  Created a content-based image retrieval system using CNNs and cosine similarity for visual comparison and clustering.
• Banner Tracker
  Built a Selenium-based scraper to track e-commerce banner visibility and behavior across platforms and apps.

LLM-powered solutions focused on generation, semantic extraction, and intelligent augmentation via prompt engineering and RAG:

• Brand and Offer Extraction
  Leveraged ChatGPT API and prompt engineering to extract brand and promotional text from marketing creatives and banners.
• Resume Ranking Tool
  Built an AI solution using LLMs and a RAG pipeline to evaluate and rank resumes based on semantic alignment with job descriptions.

• Hindi, Native
• English, Professional