Summary
Overview
Work History
Education
Skills
Certification
Timeline
Generic

Sanket Shetti

Generative AI Developer
Bengaluru,Karnataka

Summary

Experienced Generative AI Developer with the expertise in developing advanced AI-driven solutions, including product recommender systems, SDK generator. Skilled in Python, LangChain, LangGraph, Multi AI Agents Prompt Engineering, Large Language Models, Natural Language Processing and various AI and Machine Learning technologies. Recognized for exceptional work in AI project implementation and driving innovation in banking and financial services.

Overview

3
3
years of professional experience
4
4
years of post-secondary education
3
3
Certifications
1
1
Language

Work History

Generative AI Developer

Cognizant Technologies Solutions
Bengaluru
09.2022 - Current
  • API and SDK Recommendation Chatbots
    Developed product recommender and SDK suggestion chatbots leveraging Retrieval-Augmented Generation (RAG) with API documentation and pre-generated SDKs. Recommended APIs and SDKs based on user queries, streamlining the integration process for developers and improving development efficiency. Used Elastic Vector Store for optimized API retrieval and enhanced relevance by experimenting with implicit and explicit query optimization. Integrated SDK generation tools as Python packages to improve scalability and chatbot performance.
  • Diagram Generation and Visualization Tool Created an automated UML generation system to produce sequence and class diagrams from user-provided descriptions. Used LLMs for code generation, parsed with Pydantic, and compiled in a Linux environment. Converted .uml files (PlantUML) and .md files (Mermaid UML) into .png images stored in Azurite Docker containers, dynamically retrieved via blob storage.
  • Conversation Tracker and User Insights Generator Built a conversation analytics platform to categorize user queries and analyze API usage trends. Stored user history in Redis with unique IDs for tracking and generated insights like unique users, token usage, sessions, and message statistics. Used Python-based calculations for real-time and historical metrics to optimize platform performance and user engagement.
  • GPT Content Quality Scoring and Optimization Developed a human-based scoring system to evaluate GPT-generated content on metrics such as succinctness, groundedness, accuracy, relevance, and coherence. Designed a testing pipeline using BERT Models to measure precision, recall, F1 Score, Bleu Score, Rouge Score Hallucination Score and other quality metrics and utilized the Bert model for re-ranking capabilities. Conducted optimization experiments with different models and hyper parameters, improving response quality and chatbot efficiency. Refined prompt engineering techniques for better contextual accuracy and relevance in bot interactions.
  • Token Cost Estimation and Sentiment Analysis Implemented a token calculator to estimate runtime costs of GPT model usage by tracking prompt, completion, and total tokens. Incorporated a feedback system using thumbs-up/thumbs-down sentiment capture, enhanced with Microsoft Text Analytics, to improve the chatbot adaptability to user preferences.

Education

Bachelor of Technology -

Reva University
Bengaluru
08.2018 - 05.2022

Skills

  • Python

  • Generative AI

  • Prompt Engineering

  • Large Language Models

  • LangChain

  • LangGraph

  • Swagger Code Gen

  • Microsoft Azure

  • Natural Language Processing

  • Docker

  • Redis

  • Azure AI Search

  • Elastic Search

  • Deep Learning

  • Machine Learning

Certification

Microsoft Certified: AI 900 - Azure AI Fundamentals.

Timeline

Generative AI Developer

Cognizant Technologies Solutions
09.2022 - Current

Bachelor of Technology -

Reva University
08.2018 - 05.2022

Similar Profiles

CREATE PROFILE
Sanket ShettiGenerative AI Developer